McCleskey v. Kemp

=== Lead Opinion ===

Justice Powell
delivered the opinion of the Court.
This case presents the question whether a complex statistical study that indicates a risk that racial considerations enter
*283
into capital sentencing determinations proves that petitioner McCleskey’s capital sentence is unconstitutional under the Eighth or Fourteenth Amendment.
H
McCleskey, a black man, was convicted of two counts of armed robbery and one count of murder in the Superior Court of Fulton County, Georgia, on October 12, 1978. McCleskey’s convictions arose out of the robbery of a furniture store and the killing of a white police officer during the course of the robbery. The evidence at trial indicated that McCleskey and three accomplices planned and carried out the robbery. All four were armed. McCleskey entered the front of the store while the other three entered the rear. McCleskey secured the front of the store by rounding up the customers and forcing them to lie face down on the floor. The other three rounded up the employees in the rear and tied them up with tape. The manager was forced at gunpoint to turn over the store receipts, his watch, and $6. During the course of the robbery, a police officer, answering a silent alarm, entered the store through the front door. As he was walking down the center aisle of the store, two shots were fired. Both struck the officer. One hit him in the face and killed him.
Several weeks later, McCleskey was arrested in connection with an unrelated offense. He confessed that he had participated in the furniture store robbery, but denied that he had shot the police officer. At trial, the State introduced evidence that at least one of the bullets that struck the officer was fired from a .38 caliber Rossi revolver. This description matched the description of the gun that McCleskey had carried diming the robbery. The State also introduced the testimony of two witnesses who had heard McCleskey admit to the shooting.
*284
The jury convicted McCleskey of murder.
1
At the penalty hearing,
2
the jury heard arguments as to the appropriate sentence. Under Georgia law, the jury could not consider imposing the death penalty unless it found beyond a reasonable doubt that the murder was accompanied by one of the statutory aggravating circumstances.
Ga. Code Ann. § 17-10-30
(c) (1982).
3
The jury in this case found two ag
*285
gravating circumstances to exist beyond a reasonable doubt: the murder was committed during the course of an armed robbery, § 17 — 10—30(b)(2); and the murder was committed upon a peace officer engaged in the performance of his duties, § 17 — 10—30(b)(8). In making its decision whether to impose the death sentence, the jury considered the mitigating and aggravating circumstances of McCleskey’s conduct. § 17-10-2(c). McCleskey offered no mitigating evidence. The jury recommended that he be sentenced to death on the murder charge and to consecutive life sentences on the armed robbery charges. The court followed the jury’s recommendation and sentenced McCleskey to death.
4
On appeal, the Supreme Court of Georgia affirmed the convictions and the sentences.
McCleskey
v.
State,
245 Ga. 108
,
263 S. E. 2d 146
(1980). This Court denied a petition for a writ of certiorari.
McCleskey
v.
Georgia,
449 U. S. 891
(1980). The Superior Court of Fulton County denied Mc-Cleskey’s extraordinary motion for a new trial. McCleskey then filed a petition for a writ of habeas corpus in the
*286
Superior Court of Butts County. After holding an eviden-tiary hearing, the Superior Court denied relief.
McCleskey
v.
Zant,
No. 4909 (Apr. 8, 1981). The Supreme Court of Georgia denied McCleskey’s application for a certificate of probable cause to appeal the Superior Court’s denial of his petition, No. 81-5523, and this Court again denied certiorari.
McCleskey
v.
Zant,
454 U. S. 1093
(1981).
McCleskey next filed a petition for a writ of habeas corpus in the Federal District Court for the Northern District of Georgia. His petition raised 18 claims, one of which was that the Georgia capital sentencing process is administered in a racially discriminatory manner in violation of the Eighth and Fourteenth Amendments to the United States Constitution. In support of his claim, McCleskey proffered a statistical study performed by Professors David C. Baldus, Charles Pulaski, and George Woodworth (the Baldus study) that purports to show a disparity in the imposition of the death sentence in Georgia based on the race of the murder victim and, to a lesser extent, the race of the defendant. The Bal-dus study is actually two sophisticated statistical studies that examine over 2,000 murder cases that occurred in Georgia during the 1970’s. The raw numbers collected by Professor Baldus indicate that defendants charged with killing white persons received the death penalty in 11% of the cases, but defendants charged with killing blacks received the death penalty in only 1% of the cases. The raw numbers also indicate a reverse racial disparity according to the race of the defendant: 4% of the black defendants received the death penalty, as opposed to 7% of the white defendants.
Baldus also divided the cases according to the combination of the race of the defendant and the race of the victim. He found that the death penalty was assessed in 22% of the cases involving black defendants and white victims; 8% of the cases involving white defendants and white victims; 1% of the cases involving black defendants and black victims; and 3% of the cases involving white defendants and black victims.
*287
Similarly, Baldus found that prosecutors sought the death penalty in 70% of the cases involving black defendants and white victims; 32% of the cases involving white defendants and white victims; 15% of the cases involving black defendants and black victims; and 19% of the cases involving white defendants and black victims.
Baldus subjected his data to an extensive analysis, taking account of 230 variables that could have explained the disparities on nonracial grounds. One of his models concludes that, even after taking account of 39 nonracial variables, defendants charged with killing white victims were 4.3 times as likely to receive a death sentence as defendants charged with killing blacks. According to this model, black defendants were 1.1 times as likely to receive a death sentence as other defendants. Thus, the Baldus study indicates that black defendants, such as McCleskey, who kill white victims have the greatest likelihood of receiving the death penalty.
5
The District Court held an extensive evidentiary hearing on McCleskey’s petition. Although it believed that McCles-key’s Eighth Amendment claim was foreclosed by the Fifth Circuit’s decision in
Spinkellink
v.
Wainwright,
578 F. 2d 582, 612-616
(1978), cert. denied,
440 U. S. 976
(1979), it nevertheless considered the Baldus study with care. It con-
*288
eluded that McCleskey’s “statistics do not demonstrate a prima facie case in support of the contention that the death penalty was imposed upon him because of his race, because of the race of the victim, or because of any Eighth Amendment concern.”
McCleskey
v.
Zant,
580 F. Supp. 338, 379
(ND Ga. 1984). As to McCleskey’s Fourteenth Amendment claim, the court found that the methodology of the Baldus study was flawed in several respects.
6
Because of these de
*289
fects, the court held that the Baldus study “fail[ed] to contribute anything of value” to McCleskey’s claim.
Id.,
at 372
(emphasis omitted). Accordingly, the court denied the petition insofar as it was based upon the Baldus study.
The Court of Appeals for the Eleventh Circuit, sitting en banc, carefully reviewed the District Court’s decision on McCleskey’s claim.
753 F. 2d 877
(1985). It assumed the validity of the study itself and addressed the merits of Mc-Cleskey’s Eighth and Fourteenth Amendment claims. That is, the court assumed that the study “showed that systematic and substantial disparities existed in the penalties imposed upon homicide defendants in Georgia based on race of the homicide victim, that the disparities existed at a less substantial rate in death sentencing based on race of defendants, and that the factors of race of the victim and defendant were at work in Fulton County.”
Id.,
at 895
. Even assuming the study’s validity, the Court of Appeals found the statistics “insufficient to demonstrate discriminatory intent or unconstitutional discrimination in the Fourteenth Amendment context, [and] insufficient to show irrationality, arbitrariness and capriciousness under any kind of Eighth Amendment analysis.”
Id.,
at 891
. The court noted:
“The very exercise of discretion means that persons exercising discretion may reach different results from exact duplicates. Assuming each result is within the range of discretion, all are correct in the eyes of the law. It would not make sense for the system to require the exercise of discretion in order to be facially constitu
*290
tional, and at the same time hold a system unconstitutional in application where that discretion achieved different results for what appear to be exact duplicates, absent the state showing the reasons for the difference. . . .
“The Baldus approach . . . would take the cases with different results on what are contended to be duplicate facts, where the differences could not be otherwise explained, and conclude that the different result was based on race alone. . . . This approach ignores the realities. . . . There are, in fact, no exact duplicates in capital crimes and capital defendants. The type of research submitted here tends to show which of the directed factors were effective, but is of restricted use in showing what undirected factors control the exercise of constitutionally required discretion.”
Id.,
at 898-899
.
The court concluded:
“Viewed broadly, it would seem that the statistical evidence presented here, assuming its validity, confirms rather than condemns the system. . . . The marginal disparity based on the race of the victim tends to support the state’s contention that the system is working far differently from the one which
Furman
[v.
Georgia,
408 U. S. 288
(1972)] condemned. In
pre-Furman
days, there was no rhyme or reason as to who got the death penalty and who did not. But now, in the vast majority of cases, the reasons for a difference are well documented. That they are not so clear in a small percentage of the cases is no reason to declare the entire system unconstitutional.”
Id.,
at 899.
The Court of Appeals affirmed the denial by the District Court of McCleskey’s petition for a writ of habeas corpus insofar as the petition was based upon the Baldus study, with three judges dissenting as to McCleskey’s claims based on
*291
the Baldus study. We granted certiorari,
478 U. S. 1019
(1986), and now affirm.
II
McCleskey’s first claim is that the Georgia capital punishment statute violates the Equal Protection Clause of the Fourteenth Amendment.
7
He argues that race has infected the administration of Georgia’s statute in two ways: persons who murder whites are more likely to be sentenced to death than persons who murder blacks, and black murderers are more likely to be sentenced to death than white murderers.
8
*292
As a black defendant who killed a white victim, McCleskey claims that the Baldus study demonstrates that he was discriminated against because of his race and because of the race of his victim. In its broadest form, McCleskey’s claim of discrimination extends to every actor in the Georgia capital sentencing process, from the prosecutor who sought the death penalty and the jury that imposed the sentence, to the State itself that enacted the capital punishment statute and allows it to remain in effect despite its allegedly discriminatory application. We agree with the Court of Appeals, and every other court that has considered such a challenge,
9
that this claim must fail.
A
Our analysis begins with the basic principle that a defendant who alleges an equal protection violation has the burden of proving “the existence of purposeful discrimination.”
Whitus
v.
Georgia,
385 U. S. 545, 550
(1967).
10
A corollary to this principle is that a criminal defendant must prove that the purposeful discrimination “had a discriminatory effect” on him.
Wayte
v.
United States,
470 U. S. 598, 608
(1985). Thus, to prevail under the Equal Protection Clause, McCles-key must prove that the decisionmakers in
his
case acted with discriminatory purpose. He offers no evidence specific to his own case that would support an inference that racial
*293
considerations played a part in his sentence. Instead, he relies solely on the Baldus study.
11
McCleskey argues that the Baldus study compels an inference that his sentence rests on purposeful discrimination. McCleskey’s claim that these statistics are sufficient proof of discrimination, without regard to the facts of a particular case, would extend to all capital cases in Georgia, at least where the victim was white and the defendant is black.
The Court has accepted statistics as proof of intent to discriminate in certain limited contexts. First, this Court has accepted statistical disparities as proof of an equal protection violation in the selection of the jury venire in a particular district. Although statistical proof normally must present a “stark” pattern to be accepted as the sole proof of discriminatory intent under the Constitution,
12
Arlington Heights
v.
*294
Metropolitan Housing Dev. Corp.,
429 U. S. 252, 266
(1977), “[bjecause of the nature of the jury-selection task, ... we have permitted a finding of constitutional violation even when the statistical pattern does not approach [such] extremes.”
Id.,
at 266, n. 13
.
13
Second, this Court has accepted statistics in the form of multiple-regression analysis to prove statutory violations under Title VII of the Civil Rights Act of 1964.
Bazemore
v.
Friday,
478 U. S. 385, 400-401
(1986) (opinion of Brennan, J., concurring in part).
But the nature of the capital sentencing decision, and the relationship of the statistics to that decision, are fundamentally different from the corresponding elements in the venire-selection or Title VII cases. Most importantly, each particular decision to impose the death penalty is made by a petit jury selected from a properly constituted venire. Each jury is unique in its composition, and the Constitution requires that its decision rest on consideration of innumerable factors that vary according to the characteristics of the individual defendant and the facts of the particular capital offense. See
Hitchcock
v.
Dugger, post,
at 398-399;
Lockett
v.
Ohio,
438 U. S. 586, 602-605
(1978) (plurality opinion of Burger, C. J.). Thus, the application of an inference drawn from the general statistics to a specific decision in a trial and sentencing simply is not comparable to the application of an inference drawn from general statistics to a specific venire-
*295
selection or Title VII case. In those cases, the statistics relate to fewer entities,
14
and fewer variables are relevant to the challenged decisions.
15
*296
Another important difference between the cases in which we have accepted statistics as proof of discriminatory intent and this case is that, in the venire-selection and Title VII contexts, the decisionmaker has an opportunity to explain the statistical disparity. See
Whitus
v.
Georgia,
385 U. S., at 552
;
Texas Dept. of Community Affairs
v.
Burdine,
450 U. S. 248, 254
(1981);
McDonnell Douglas Corp.
v.
Green,
411 U. S. 792, 802
(1973). Here, the State has no practical opportunity to rebut the Baldus study. “[Cjontrolling considerations of . . . public policy,”
McDonald
v.
Pless,
238 U. S. 264, 267
(1915), dictate that jurors “cannot be called ... to testify to the motives and influences that led to their verdict.”
Chicago, B. & Q. R. Co.
v.
Babcock,
204 U. S. 585, 593
(1907). Similarly, the policy considerations behind a prosecutor’s traditionally “wide discretion”
16
suggest the impropriety of our requiring prosecutors to defend their decisions to seek death penalties, “often years after they were made.”
17
See
Imbler
v.
Pachtman,
424 U. S. 409, 425-426
(1976).
18
Moreover, absent far stronger proof, it is unnec
*297
essary to seek such a rebuttal, because a legitimate and unchallenged explanation for the decision is apparent from the record: McCleskey committed an act for which the United States Constitution and Georgia laws permit imposition of the death penalty.
19
Finally, McCleskey’s statistical proffer must be viewed in the context of his challenge. McCleskey challenges decisions at the heart of the State’s criminal justice system. “[0]ne of society’s most basic tasks is that of protecting the lives of its citizens and one of the most basic ways in which it achieves the task is through criminal laws against murder.”
Gregg
v.
Georgia,
428 U. S. 153, 226
(1976) (White, J., concurring). Implementation of these laws necessarily requires discretionary judgments. Because discretion is essential to the criminal justice process, we would demand exceptionally clear proof before we would infer that the discretion has been abused. The unique nature of the decisions at issue in this case also counsels against adopting such an inference from the disparities indicated by the Baldus study. Accordingly, we hold that the Baldus study is clearly insufficient to support an inference that any of the decisionmakers in McCles-key’s case acted with discriminatory purpose.
B
McCleskey also suggests that the Baldus study proves that the State as a whole has acted with a discriminatory purpose. He appears to argue that the State has violated the Equal
*298
Protection Clause by adopting the capital punishment statute and allowing it to remain in force despite its allegedly discriminatory application. But “ ‘[discriminatory purpose’... implies more than intent as volition or intent as awareness of consequences. It implies that the decisionmaker, in this case a state legislature, selected or reaffirmed a particular course of action at least in part ‘because of,’ not merely ‘in spite of,’ its adverse effects upon an identifiable group.”
Personnel Administrator of Massachusetts
v.
Feeney,
442 U. S. 256, 279
(1979) (footnote and citation omitted). See
Wayte
v.
United States,
470 U. S., at 608-609
. For this claim to prevail, McCleskey would have to prove that the Georgia Legislature enacted or maintained the death penalty statute
because of
an anticipated racially discriminatory effect. In
Gregg
v.
Georgia, supra,
this Court found that the Georgia capital sentencing system could operate in a fair and neutral manner. There was no evidence then, and there is none now, that the Georgia Legislature enacted the capital punishment statute to further a racially discriminatory purpose.
20
Nor has McCleskey demonstrated that the legislature maintains the capital punishment statute because of the racially disproportionate impact suggested by the Baldus study. As legislatures necessarily have wide discretion in the choice of criminal laws and penalties, and as there were
*299
legitimate reasons for the Georgia Legislature to adopt and maintain capital punishment, see
Gregg
v.
Georgia, supra,
at 183-187 (joint opinion of Stewart, Powell, and Stevens, JJ.), we will not infer a discriminatory purpose on the part of the State of Georgia.
21
Accordingly, we reject McCleskey’s equal protection claims.
Ill
McCleskey also argues that the Baldus study demonstrates that the Georgia capital sentencing system violates the Eighth Amendment.
22
We begin our analysis of this claim by reviewing the restrictions on death sentences established by our prior decisions under that Amendment.
A
The Eighth Amendment prohibits infliction of “cruel and unusual punishments.” This Court’s early Eighth Amendment cases examined only the “particular methods of execution to determine whether they were too cruel to pass constitutional muster.”
Gregg
v.
Georgia, supra,
at 170. See
In re Kemmler,
136 U. S. 436
(1890) (electrocution);
*300
Wilkerson
v.
Utah,
99 U. S. 130
(1879) (public shooting). Subsequently, the Court recognized that the constitutional prohibition against cruel and unusual punishments “is not fastened to the obsolete but may acquire meaning as public opinion becomes enlightened by a humane justice.”
Weems
v.
United States,
217 U. S. 349, 378
(1910). In
Weems
,
the Court identified a second principle inherent in the Eighth Amendment, “that punishment for crime should be graduated and proportioned to offense.”
Id.,
at 367
.
Chief Justice Warren, writing for the plurality in
Trop
v.
Dulles,
356 U. S. 86, 99
(1958), acknowledged the constitutionality of capital punishment. In his view, the “basic concept underlying the Eighth Amendment” in this area is that the penalty must accord with “the dignity of man. ”
Id.,
at 100
. In applying this mandate, we have been guided by his statement that “[t]he Amendment must draw its meaning from the evolving standards of decency that mark the progress of a maturing society.”
Id.,
at 101
. Thus, our constitutional decisions have been informed by “contemporary values concerning the infliction of a challenged sanction,”
Gregg
v.
Georgia,
428 U. S., at 173
. In assessing contemporary values, we have eschewed subjective judgment, and instead have sought to ascertain “objective indicia that reflect the public attitude toward a given sanction.”
Ibid.
First among these indicia are the decisions of state legislatures, “because the . . . legislative judgment weighs heavily in ascertaining” contemporary standards,
id.,
at 175
. We also have been guided by the sentencing decisions of juries, because they are “a significant and reliable objective index of contemporary values,”
id.,
at 181
. Most of our recent decisions as to the constitutionality of the death penalty for a particular crime have rested on such an examination of contemporary values.
E. g., Enmund
v.
Florida,
458 U. S. 782, 789-796
(1982) (felony murder);
Coker
v.
Georgia,
433 U. S. 584, 592-597
(1977) (plurality opinion of White, J.) (rape);
Gregg
v.
Georgia, supra,
at 179-182 (murder).
*301
B
Two principal decisions guide our resolution of McCles-key’s Eighth Amendment claim. In
Furman
v.
Georgia,
408 U. S. 238
(1972), the Court concluded that the death penalty was so irrationally imposed that any particular death sentence could be presumed excessive. Under the statutes at issue in
Furman,
there was no basis for determining in any particular case whether the penalty was proportionate to the crime: “[T]he death penalty [was] exacted with great infrequency even for the most atrocious crimes and . . . there [was] no meaningful basis for distinguishing the few cases in which it [was] imposed from the many cases in which it [was] not.”
Id.,
at 313 (White, J., concurring).
In
Gregg
,
the Court specifically addressed the question left open in
Furman
— whether the punishment of death for murder is “under all circumstances, ‘cruel and unusual’ in violation of the Eighth and Fourteenth Amendments of the Constitution.”
428 U. S., at 168
. We noted that the imposition of the death penalty for the crime of murder “has a long history of acceptance both in the United States and in England.”
Id.,
at 176
(joint opinion of Stewart, Powell, and Stevens, JJ.). “The most marked indication of society’s endorsement of the death penalty for murder [was] the legislative response to
Furman.” Id.,
at 179
. During the 4-year period between
Furman
and
Gregg
,
at least 35 States had reenacted the death penalty, and Congress had authorized the penalty for aircraft piracy.
428 U. S., at 179-180
.
23
The “actions of juries” were “fully compatible with the legislative judgments.”
Id.,
at 182
. We noted that any punishment might be unconstitutionally severe if inflicted without peno-logical justification, but concluded:
*302
“Considerations of federalism, as well as respect for the ability of a legislature to evaluate, in terms of its particular State, the moral consensus concerning the death penalty and its social utility as a sanction, require us to conclude, in the absence of more convincing evidence, that the infliction of death as a punishment for murder is not without justification and thus is not unconstitutionally severe.”
Id.,
at 186-187
.
The second question before the Court in
Gregg
was the constitutionality of the particular procedures embodied in the Georgia capital punishment statute. We explained the fundamental principle of
Furman,
that “where discretion is afforded a sentencing body on a matter so grave as the determination of whether a human life should be taken or spared, that discretion must be suitably directed and limited so as to minimize the risk of wholly arbitrary and capricious action.”
428 U. S., at 189
. Numerous features of the then new Georr gia statute met the concerns articulated in Furman.
24
The Georgia system bifurcates guilt and sentencing proceedings so that the jury can receive all relevant information for sentencing without the risk that evidence irrelevant to the defendant’s guilt will influence the jury’s consideration of that issue. The statute narrows the class of murders subject to the death penalty to cases in which the jury finds at least one statutory aggravating circumstance beyond a reasonable doubt. Conversely, it allows the defendant to introduce any relevant mitigating evidence that might influence the jury not to impose a death sentence. See
428 U. S., at 163-164
. The procedures also require a particularized inquiry into “ ‘the circumstances of the offense together with the character and propensities of the offender.’”
Id.,
at 189 (quoting
Pennsylvania ex rel. Sullivan
v.
Ashe,
302 U. S. 51, 55
(1937)). Thus, “while some jury discretion still exists, ‘the
*303
discretion to be exercised is controlled by clear and objective standards so as to produce non-discriminatory application.’”
428 U. S., at 197
-198 (quoting
Coley
v.
State,
231 Ga. 829, 834
,
204 S. E. 2d 612, 615
(1974)). Moreover, the Georgia system adds “an important additional safeguard against arbitrariness and caprice” in a provision for automatic appeal of a death sentence to the State Supreme Court.
428 U. S., at 198
. The statute requires that court to review each sentence to determine whether it was imposed under the influence of passion or prejudice, whether the evidence supports the jury’s finding of a statutory aggravating circumstance, and whether the sentence is disproportionate to sentences imposed in generally similar murder cases. To aid the court’s review, the trial judge answers a questionnaire about the trial, including detailed questions as to “the quality of the defendant’s representation [and] whether race played a role in the trial.”
Id.,
at 167
.
C
In the cases decided after
Gregg
,
the Court has imposed a number of requirements on the capital sentencing process to ensure that capital sentencing decisions rest on the individualized inquiry contemplated in
Gregg
.
In
Woodson
v.
North Carolina,
428 U. S. 280
(1976), we invalidated a mandatory capital sentencing system, finding that the “respect for humanity underlying the Eighth Amendment requires consideration of the character and record of the individual offender and the circumstances of the particular offense as a constitutionally indispensable part of the process of inflicting the penalty of death.”
Id.,
at 304
(plurality opinion of Stewart, Powell, and Stevens, JJ.) (citation omitted). Similarly, a State must “narrow the class of murderers subject to capital punishment,”
Gregg
v.
Georgia, supra,
at 196, by providing “specific and detailed guidance” to the sentencer.
25
*304
Proffitt
v.
Florida,
428 U. S. 242, 253
(1976) (joint opinion of Stewart, Powell, and Stevens, JJ.).
In contrast to the carefully defined standards that must narrow a sentencer’s discretion to
impose
the death sentence, the Constitution limits a State’s ability to narrow a sen-tencer’s discretion to consider relevant evidence that might cause it to
decline to impose
the death sentence.
26
“[T]he sentencer . . . [cannot] be precluded from considering,
as a mitigating factor,
any aspect of a defendant’s character or recorcl and any of the circumstances of the offense that the defendant proffers as a basis for a sentence less than death.”
Lockett
v.
Ohio,
438 U. S., at 604
(plurality opinion of Burger, C. J.) (emphasis in original; footnote omitted). See
Skipper
v.
South Carolina,
476 U. S. 1
(1986). Any exclusion of the “compassionate or mitigating factors stemming from the diverse frailties of humankind” that are relevant to the sentencer’s decision would fail to treat all persons as “uniquely individual human beings.”
Woodson
v.
North Carolina, supra,
at 304.
Although our constitutional inquiry has centered on the procedures by which a death sentence is imposed, we have not stopped at the face of a statute, but have probed the appli
*305
cation of statutes to particular cases. For example, in
Godfrey
v.
Georgia,
446 U. S. 420
(1980), the Court invalidated a Georgia Supreme Court interpretation of the statutory aggravating circumstance that the murder be “outrageously or wantonly vile, horrible or inhuman in that it involved torture, depravity of mind, or an aggravated battery to the victim.” Ga. Code § 27-2534.1(b)(7) (1978).
27
Although that court had articulated an adequate limiting definition of this phrase, we concluded that its interpretation in
Godfrey
was so broad that it may have vitiated the role of the aggravating circumstance in guiding the sentencing jury’s discretion.
Finally, where the objective indicia of community values have demonstrated a consensus that the death penalty is disproportionate as applied to a certain class of cases, we have established substantive limitations on its application. In
Coker
v.
Georgia,
433 U. S. 584
(1977), the Court held that a State may not constitutionally sentence an individual to death for the rape of an adult woman. In
Enmund
v.
Florida,
458 U. S. 782
(1982), the Court prohibited imposition of the death penalty on a defendant convicted of felony murder absent a showing that the defendant possessed a sufficiently culpable mental state. Most recently, in
Ford
v.
Wainwright,
477 U. S. 399
(1986), we prohibited execution of prisoners who are insane.
D
In sum, our decisions since
Furman
have identified a constitutionally permissible range of discretion in imposing the death penalty. First, there is a required threshold below which the death penalty cannot be imposed. In this context, the State must establish rational criteria that narrow the decisionmaker’s judgment as to whether the circumstances of a particular defendant’s case meet the threshold. Moreover, a societal consensus that the death penalty is disproportion
*306
ate to a particular offense prevents a State from imposing the death penalty for that offense. Second, States cannot limit the sentencer’s consideration of any relevant circumstance that could cause it to decline to impose the penalty. In this respect, the State cannot channel the sentencer’s discretion, but must allow it to consider any relevant information offered by the defendant.
IV
A
In light of our precedents under the Eighth Amendment, McCleskey cannot argue successfully that his sentence is “disproportionate to the crime in the traditional sense.” See
Pulley
v.
Harris,
465 U. S. 37, 43
(1984). He does not deny that he committed a murder in the course of a planned robbery, a crime for which this Court has determined that the death penalty constitutionally may be imposed.
Gregg
v.
Georgia,
428 U. S., at 187
. His disproportionality claim “is of a different sort.”
Pulley
v.
Harris, supra,
at 43
. Mc-Cleskey argues that the sentence in his case is disproportionate to the sentences in other murder cases.
On the one hand, he cannot base a constitutional claim on an argument that his case differs from other cases in which defendants
did
receive the death penalty. On automatic appeal, the Georgia Supreme Court found that McCleskey’s death sentence was not disproportionate to other death sentences imposed in the State.
McCleskey
v.
State,
245 Ga. 108
,
263 S. E. 2d 146
(1980). The court supported this conclusion with an appendix containing citations to 13 cases involving generally similar murders. See Ga. Code Ann; § 17-10-35(e) (1982). Moreover, where the statutory procedures adequately channel the sentencer’s discretion, such proportionality review is not constitutionally required.
Pulley
v.
Harris, supra,
at 50-51
.
On the other hand, absent a showing that the Georgia capital punishment system operates in an arbitrary and capricious manner, McCleskey cannot prove a constitutional
*307
violation by demonstrating that other defendants who may be similarly situated did
not
receive the death penalty. In
Gregg
,
the Court confronted the argument that “the opportunities for discretionary action that are inherent in the processing of any murder case under Georgia law,” 428 U. S., at 199, specifically the opportunities for discretionary leniency, rendered the capital sentences imposed arbitrary and capricious. We rejected this contention:
“The existence of these discretionary stages is not determinative of the issues before us. At each of these stages an actor in the criminal justice system makes a decision which may remove a defendant from consideration as a candidate for the death penalty.
Furman,
in contrast, dealt with the decision to impose the death sentence on a specific individual who had been convicted of a capital offense. Nothing in any of our cases suggests that the decision to afford an individual defendant mercy violates the Constitution.
Furman
held only that, in order to minimize the risk that the death penalty would be imposed on a capriciously selected group of offenders, the decision to impose it had to be guided by standards so that the sentencing authority would focus on the particularized circumstances of the crime and the defendant.”
Ibid,.
28
*308
Because McCleskey’s sentence was imposed under Georgia sentencing procedures that focus discretion “on the particularized nature of the crime and the particularized characteristics of the individual defendant,”
id.,
at 206, we lawfully may presume that McCleskey’s death sentence was not “wantonly and freakishly” imposed,
id.,
at 207, and thus that the sentence is not disproportionate within any recognized meaning under the Eighth Amendment.
B
Although our decision in
Gregg
as to the facial validity of the Georgia capital punishment statute appears to foreclose McCleskey’s disproportionality argument, he further contends that the Georgia capital punishment system is arbitrary and capricious in
application,
and therefore his sentence is excessive, because racial considerations may influence capital sentencing decisions in Georgia. We now address this claim.
To evaluate McCleskey’s challenge, we must examine exactly what the Baldus study may show. Even Professor Bal-dus does not contend that his statistics
prove
that race enters into any capital sentencing decisions or that race was a factor in McCleskey’s particular case.
29
Statistics at most may show only a likelihood that a particular factor entered into some decisions. There is, of course, some risk of racial prejudice influencing a jury’s decision in a criminal case. There are similar risks that other kinds of prejudice will influence other criminal trials. See
infra,
at 315-318. The question
*309
“is at what point that risk becomes constitutionally unacceptable,”
Turner
v.
Murray,
476 U. S. 28, 36, n. 8
(1986). McCleskey asks us to accept the likelihood allegedly shown by the Baldus study as the constitutional measure of an unacceptable risk of racial prejudice influencing capital sentencing decisions. This we decline to do.
Because of the risk that the factor of race may enter the criminal justice process, we have engaged in “unceasing efforts” to eradicate racial prejudice from our criminal justice system.
Batson
v.
Kentucky,
476 U. S. 79, 85
(1986).
30
Our efforts have been guided by our recognition that “the inestimable privilege of trial by jury ... is a vital principle, underlying the whole administration of criminal justice,”
Ex parte Milligan,
4 Wall. 2
, 123 (1866). See
Duncan
v.
*310
Louisiana,
391 U. S. 145, 155
(1968).
31
Thus, it is the jury that is a criminal defendant’s fundamental “protection of life and liberty against race or color prejudice.”
Strauder
v.
West Virginia,
100 U. S. 303, 309
(1880). Specifically, a capital sentencing jury representative of a criminal defendant’s community assures a “‘diffused impartiality,’”
Taylor
v.
Louisiana,
419 U. S. 522, 530
(1975) (quoting
Thiel
v.
Southern Pacific Co.,
328 U. S. 217, 227
(1946) (Frankfurter, J., dissenting)), in the jury’s task of “express[ing] the conscience of the community on the ultimate question of life or death,”
Witherspoon
v.
Illinois,
391 U. S. 510, 519
(1968).
32
*311
Individual jurors bring to their deliberations “qualities of human nature and varieties of human experience, the range of which is unknown and perhaps unknowable.”
Peters
v.
Kiff,
407 U. S. 493, 503
(1972) (opinion of Marshall, J.). The capital sentencing decision requires the individual jurors to focus their collective judgment on the unique characteristics of a particular criminal defendant. It is not surprising that such collective judgments often are difficult to explain. But the inherent lack of predictability of jury decisions does not justify their condemnation. On the contrary, it is the jury’s function to make the difficult and uniquely human judgments that defy codification and that “buil[d] discretion, equity, and flexibility into a legal system.” H. Kalven & H. Zeisel, The American Jury 498 (1966).
McCleskey’s argument that the Constitution condemns the discretion allowed decisionmakers in the Georgia capital sentencing system is antithetical to the fundamental role of discretion in our criminal justice system. Discretion in the criminal justice system offers substantial benefits to the criminal defendant. Not only can a jury decline to impose the death sentence, it can decline to convict or choose to convict of a lesser offense. Whereas decisions against a defendant’s interest may be reversed by the trial judge or on appeal, these discretionary exercises of leniency are final and unreviewable.
33
Similarly, the capacity of prosecutorial dis
*312
cretion to provide individualized justice is “firmly entrenched in American law.” 2 W. LaFave & J. Israel, Criminal Procedure § 13.2(a), p. 160 (1984). As we have noted, a prosecutor can decline to charge, offer a plea bargain,
34
or decline to seek a death sentence in any particular case. See n. 28,
supra.
Of course, “the power to be lenient [also] is the power to discriminate,” K. Davis, Discretionary Justice 170 (1973), but a capital punishment system that did not allow for discretionary acts of leniency “would be totally alien to our notions of criminal justice.”
Gregg
v.
Georgia,
428 U. S., at 200, n. 50
.
C
At most, the Baldus study indicates a discrepancy that appears to correlate with race. Apparent disparities in sentencing are an inevitable part of our criminal justice system.
35
*313
The discrepancy indicated by the Baldus study is “a far cry from the major systemic defects identified in
Furman,” Pulley v
.
Harris,
465 U. S., at 54
.
36
As this Court has recognized, any mode for determining guilt or punishment “has its weaknesses and the potential for misuse.”
Singer
v.
United States,
380 U. S. 24, 35
(1965). See
Bordenkircher
v.
Hayes,
434 U. S. 357, 365
(1978). Specifically, “there can be ‘no perfect procedure for deciding in which cases governmental authority should be used to impose death.’ ”
Zant
v.
Stephens,
462 U. S. 862, 884
(1983) (quoting
Lockett
v.
Ohio,
438 U. S., at 605
(plurality opinion of Burger, C. J.)). Despite these imperfections, our consistent rule has been that constitutional guarantees are met when “the mode [for determining guilt or punishment] itself has been surrounded with safeguards to make it as fair as possible.”
Singer
v.
United States, supra,
at 35. Where the discretion that is fundamental to our criminal process is involved, we decline to assume that what is unexplained is invidious. In light of the safeguards designed to minimize racial bias in the process, the fundamental value of jury trial in our criminal justice system, and the benefits that discretion provides to criminal defendants, we hold that the Baldus study does not demonstrate a constitutionally significant risk of racial bias affecting the Georgia capital sentencing process.
37
*314
V
Two additional concerns inform our decision in this case. First, McCleskey’s claim, taken to its logical conclusion,
*315
throws into serious question the principles that underlie our entire criminal justice system. The Eighth Amendment is not limited in application to capital punishment, but applies to all penalties.
Solem
v.
Helm,
463 U. S. 277, 289-290
(1983); see
Rummel
v.
Estelle,
445 U. S. 263, 293
(1980) (Powell, J., dissenting). Thus, if we accepted McCleskey’s claim that racial bias has impermissibly tainted the capital sentencing decision, we could soon be faced with similar claims as to other types of penalty.
38
Moreover, the claim that his sen
*316
tence rests on the irrelevant factor of race easily could be extended to apply to claims based on unexplained discrepancies that correlate to membership in other minority groups,
39
and
*317
even to gender.
40
Similarly, since McCleskey’s claim relates to the race of his victim, other claims could apply with equally logical force to statistical disparities that correlate with the race or sex of other actors in the criminal justice system, such as defense attorneys
41
or judges.
42
Also, there is no logical reason that such a claim need be limited to racial or sexual bias. If arbitrary and capricious punishment is the touchstone under the Eighth Amendment, such a claim could — at least in theory — be based upon any arbitrary variable, such as the defendant’s facial characteristics,
43
or the physical attractiveness of the defendant or the victim,
44
that some sta
*318
tistical study indicates may be influential in jury decision-making. As these examples illustrate, there is no limiting principle to the type of challenge brought by McCleskey.
45
*319
The Constitution does not require that a State eliminate any demonstrable disparity that correlates with a potentially irrelevant factor in order to operate a criminal justice system that includes capital punishment. As we have stated specifically in the context of capital punishment, the Constitution does not “plac[e] totally unrealistic conditions on its use.”
Gregg
v.
Georgia,
428 U. S., at 199, n. 50
.
Second, McCleskey’s arguments are best presented to the legislative bodies. It is not the responsibility — or indeed even the right — of this Court to determine the appropriate punishment for particular crimes. It is the legislatures, the elected representatives of the people, that are “constituted to respond to the will and consequently the moral values of the people.”
Furman
v.
Georgia,
408 U. S., at 383
(Burger, C. J., dissenting). Legislatures also are better qualified to weigh and “evaluate the results of statistical studies in terms of their own local conditions and with a flexibility of approach that is not available to the courts,”
Gregg
v.
Georgia, supra,
at 186. Capital punishment is now the law in more than two-thirds of our States. It is the ultimate duty of courts to determine on a case-by-case basis whether these laws are applied consistently with the Constitution. Despite McCleskey’s wide-ranging arguments that basically challenge the validity of capital punishment in our multiracial society, the only question before us is whether in his case, see
supra,
at 283-285, the law of Georgia was properly applied. We agree with the District Court and the Court of Appeals for the Eleventh Circuit that this was carefully and correctly done in this case.
*320
<1 HH
Accordingly, we affirm the judgment of the Court of Appeals for the Eleventh Circuit.
It is so ordered.
The Georgia Code has been revised and renumbered since McCleskey’s trial. The changes do not alter the substance of the sections relevant to this case. For convenience, references in this opinion are to the current sections.
The Georgia Code contains only one degree of murder. A person commits murder “when he unlawfully and with malice aforethought, either express or implied, causes the death of another human being.”
Ga. Code Ann. § 16-5
-l(a) (1984). A person convicted of murder “shall be punished by death or by imprisonment for life.” § 16 — 5—1(d).
Georgia Code Ann. § 17-10-2(c) (1982) provides that when a jury convicts a defendant of murder, “the court shall resume the trial and conduct a presentence hearing before the jury.” This subsection suggests that a defendant convicted of murder always is subjected to a penalty hearing at which the jury considers imposing a death sentence. But as a matter of practice, penalty hearings seem to be held only if the prosecutor affirmatively seeks the death penalty. If he does not, the defendant receives a sentence of life imprisonment. See Baldus, Pulaski, & Woodworth, Comparative Review of Death Sentences: An Empirical Study of the Georgia Experience, 74 J. Crim. L. & C. 661, 674, n. 56 (1983).
A jury cannot sentence a defendant to death for murder unless it finds that one of the following aggravating circumstances exists beyond a reasonable doubt:
“(1) The offense . . . was committed by a person with a prior record of conviction for a capital felony;
“(2) The offense . . . was committed while the offender was engaged in the commission of another capital felony or aggravated battery, or the offense of murder was committed while the offender was engaged in the commission of burglary or arson in the first degree;
“(3) The offender, by his act of murder. . . knowingly created a great risk of death to more than one person in a public place by means of a weapon or device which would normally be hazardous to the lives of more than one person;
“(4) The offender committed the offense ... for himself or another, for the purpose of receiving money or any other thing of monetary value;
*285
“(5) The murder of a judicial officer, former judicial officer, district attorney or solicitor, or former district attorney or solicitor was commmitted during or because of the exercise of his official duties;
“(6) The offender caused or directed another to commit murder or committed murder as an agent or employee of another person;
“(7) The offense of murder, rape, armed robbery, or kidnapping was outrageously or wantonly vile, horrible, or inhuman in that it involved torture, depravity of mind, or an aggravated battery to the victim;
“(8) The offense . . . was committed against any peace officer, corrections employee, or fireman while engaged in the performance of his official duties;
“(9) The offense . . . was committed by a person in, or who has escaped from, the lawful custody of a peace officer or place of lawful confinement; or
“(10) The murder was committed for the purpose of avoiding, interfering with, or preventing a lawful arrest or custody in a place of lawful confinement, of himself or another.” § 17-10-30(b).
Georgia law provides that “[wjhere a statutory aggravating circumstance is found and a recommendation of death is made, the court shall sentence the defendant to death.” § 17-10-31.
Baldus’ 230-variable model divided eases into eight different ranges, according to the estimated aggravation level of the offense. Baldus argued in his testimony to the District Court that the effects of racial bias were most striking in the midrange cases. “[W]hen the cases become tremendously aggravated so that everybody would agree that if we’re going to have a death sentence, these are the cases that should get it, the race effects go away. It’s only in the mid-range of cases where the decision-makers have a real choice as to what to do. If there’s room for the exercise of discretion, then the [racial] factors begin to play a role.” App. 36. Under this model, Baldus found that 14.4% of the black-victim midrange eases received the death penalty, and 34.4% of the white-victim cases received the death penalty. See Exhibit DB 90, reprinted in Supplemental Exhibits 64. According to Baldus, the facts of M’cCleskey’s case placed it within the midrange. App. 45-46.
Baldus, among other experts, testified at the evidentiary hearing. The District Court “was impressed with the learning of all of the experts.”
580 F. Supp., at 353
(emphasis omitted). Nevertheless, the District Court noted that in many respects the data were incomplete. In its view, the questionnaires used to obtain the data failed to capture the full degree of the aggravating or mitigating circumstances.
Id., at 356
. The court criticized the researcher’s decisions regarding unknown variables.
Id.,
at 357-358
. The researchers could not discover whether penalty trials were held in many of the cases, thus undercutting the value of the study’s statistics as to prosecutorial decisions.
Id.,
at 359
. In certain cases, the study lacked information on the race of the victim in cases involving multiple victims, on whether or not the prosecutor offered a plea bargain, and on credibility problems with witnesses.
Id.,
at 360
. The court concluded that McCleskey had failed to establish by a preponderance of the evidence that the data were trustworthy. “It is a major premise of a statistical case that the data base numerically mirrors reality. If it does not in substantial degree mirror reality, any inferences empirically arrived at are untrustworthy.”
Ibid.
The District Court noted other problems with Baldus’ methodology. First, the researchers assumed that all of the information available from the questionnaires was available to the juries and prosecutors when the case was tried. The court found this assumption “questionable.”
Id.,
at 361
. Second, the court noted the instability of the various models. Even with the 230-variable model, consideration of 20 further variables caused a significant drop in the statistical significance of race. In the court’s view, this undermined the persuasiveness of the model that showed the greatest racial disparity, the 39-variable model.
Id.,
at 362
. Third, the court found that the high correlation between race and many of the nonracial variables diminished the weight to which the study was entitled.
Id.,
at 363-364
.
Finally, the District Court noted the inability of any of the models to predict the outcome of actual cases. As the court explained, statisticians use
*289
a measure called an “r2” to measure what portion of the variance in the dependent variable (death sentencing rate, in this case) is accounted for by the independent variables of the model. A perfectly predictive model would have an r2 value of 1.0. A model with no predictive power would have an r2 value of 0. The r2 value of Baldus’ most complex model, the 230-variable model, was between .46 and .48. Thus, as the court explained, “the 230-variable model does not predict the outcome in half of the eases.”
Id.,
at 361
.
Although the District Court rejected the findings of the Baldus study as flawed, the Court of Appeals assumed that the study is valid and reached the constitutional issues. Accordingly, those issues are before us. As did the Court of Appeals, we assume the study is valid statistically without reviewing the factual findings of the District Court. Our assumption that the Baldus study is statistically valid does not include the assumption that the study shows that racial considerations actually enter into any sentencing decisions in Georgia. Even a sophisticated multiple-regression analysis such as the Baldus study can only demonstrate a
risk
that the factor of race entered into some capital sentencing decisions and a necessarily lesser risk that race entered into any particular sentencing decision.
Although McCleskey has standing to claim that he suffers discrimination because of his own race, the State argues that he has no standing to contend that he was discriminated against on the basis of his victim’s race. While it is true that we are reluctant to recognize “standing to assert the rights of third persons,”
Arlington Heights
v.
Metropolitan Housing Dev. Corp.,
429 U. S. 252, 263
(1977), this does not appear to be the nature of McCleskey’s claim. He does not seek to assert some right of his victim, or the rights of black murder victims in general. Rather, McCleskey argues that application of the State’s statute has created a classification that is “an irrational exercise of governmental power,” Brief for Petitioner 41, because it is not “necessary to the accomplishment of some permissible state objective.”
Loving
v.
Virginia,
388 U. S. 1, 11
(1967). See
McGowan
v.
Maryland,
366 U. S. 420, 425
(1961) (statutory classification cannot be “wholly irrelevant to the achievement of the State’s objective”). ' It would violate the Equal Protection Clause for a State to base enforcement of its criminal laws on “an unjustifiable standard such as race, religion, or other arbitrary classification.”
Oyler
v.
Boles,
368 U. S. 448, 456
(1962). See
*292
Cleveland Bd. of Ed.
v.
Lafleur,
414 U. S. 632, 652-653
(1974) (Powell, J., concurring). Because McCleskey raises such a claim, he has standing.
See,
e. g., Shaw
v.
Martin,
733 F. 2d 304, 311-314
(CA4), cert. denied,
469 U. S. 873
(1984);
Adams
v.
Wainwright,
709 F. 2d 1443
(CA11 1983)
(per curiam),
cert. denied,
464 U. S. 1063
(1984);
Smith
v.
Balkcom,
660 F. 2d 573, 584-585
, modified,
671 F. 2d 858, 859-860
(CA5 Unit B 1981)
(per curiam),
cert. denied,
459 U. S. 882
(1982);
Spinkellink
v.
Wainwright,
578 F. 2d 582, 612-616
(CA5 1978), cert. denied,
440 U. S. 976
(1979).
See
Arlington Heights
v.
Metropolitan Housing Dev. Corp., supra,
at 265;
Washington
v.
Davis,
426 U. S. 229, 240
(1976).
McCleskey’s expert testified:
“Models that are developed talk about the effect on the average. They do not depict the experience of a single individual. What they say, for example, [is] that on the average, the race of the victim, if it is white, increases on the average the probability . . . (that) the death sentence would be given.
“Whether in a given case that is the answer, it cannot be determined from statistics.”
580 F. Supp., at 372
.
Gomillion
v.
Lightfoot,
364 U. S. 339
(1960), and
Yick Wo
v.
Hopkins,
118 U. S. 356
(1886), are examples of those rare cases in which a statistical pattern of discriminatory impact demonstrated a constitutional violation. In
Gomülion,
a state legislature violated the Fifteenth Amendment by altering the boundaries of a particular city “from a square to an uncouth twenty-eight-sided figure.”
364 U. S., at 340
. The alterations excluded 395 of 400 black voters without excluding a single white voter. In
Yick Wo
,
an ordinance prohibited operation of 310 laundries that were housed in wooden buildings, but allowed such laundries to resume operations if the operator secured a permit from the government. When laundry operators applied for permits to resume operation, all but one of the white applicants received- permits, but none of the over 200 Chinese applicants were successful. In those cases, the Court found the statistical disparities “to warrant and require,”
Yick Wo
v.
Hopkins, supra,
at 373
, a “conclusion [that
*294
was] irresistible, tantamount for all practical purposes to a mathematical demonstration,”
Gomillion
v.
Lightfoot, supra,
at 341
, that the State acted with a discriminatory purpose.
See,
e. g., Castaneda
v.
Partida,
430 U. S. 482, 495
(1977) (2-to-1 disparity between Mexican-Americans in county population and those summoned for grand jury duty);
Turner
v.
Fouche,
396 U. S. 346, 359
(1970) (1.6-to-1 disparity between blacks in county population and those on grand jury lists);
Whitus
v.
Georgia,
385 U. S. 545, 552
(1967) (3-to-l disparity between eligible blacks in county and blacks on grand jury venire).
In venire-selection cases, the factors that may be considered are limited, usually by state statute. See
Castaneda
v.
Partida, supra,
at 485
(“A grand juror must be a citizen of Texas and of the county, be a qualified voter in the county, be ‘of sound mind and good moral character,’ be literate, have no prior felony conviction, and be under no pending indictment ‘or other legal accusation for theft or of any felony’ ”);
Turner
v.
Fouche, supra,
at 354
(jury commissioners may exclude any not “upright” and “intelligent” from grand jury service);
Whitus
v.
Georgia, supra,
at 548 (same). These considerations are uniform for all potential jurors, and although some factors may be said to be subjective, they are limited and, to a great degree, objectively verifiable. While employment decisions may involve a number of relevant variables, these variables are to a great extent uniform for all employees because they must all have a reasonable relationship to the employee’s qualifications to perform the particular job at issue. Identifiable qualifications for a single job provide a common standard by which to assess each employee. In contrast, a capital sentencing jury may consider
any
factor relevant to the defendant’s background, character, and the offense. See
Eddings
v.
Oklahoma,
455 U. S. 104, 112
(1982). There is no common standard by which to evaluate all defendants who have or have not received the death penalty.
We refer here not to the number of entities involved in any particular decision, but to the number of entities whose decisions necessarily'are reflected in a statistical display such as the Baldus study. The decisions of a jury commission or of an employer over time are fairly attributable to the commission or the employer. Therefore, an unexplained statistical discrepancy can be said to indicate a consistent policy of the decisionmaker. The Baldus study seeks to deduce a state “policy” by studying the combined effects of the decisions of hundreds of juries that are unique in their composition. It is incomparably more difficult to deduce a consistent policy by studying the decisions of these many unique entities. It is also questionable whether any consistent policy can be derived by studying the decisions of prosecutors. The District Attorney is elected by the voters in a particular county. See Ga. Const., Art. 6, §8, ¶1. Since decisions whether to prosecute and what to charge necessarily are individualized and involve infinite factual variations, coordination among district attorney offices across a State would be relatively meaningless. Thus, any inference from statewide statistics to a prosecutorial “policy” is of doubtful rele
*296
vance. Moreover, the statistics in Pulton County alone represent the disposition of far fewer cases than the statewide statistics. Even assuming the statistical validity of the Baldus study as a whole, the weight to be given the results gleaned from this small sample is limited.
See
Wayte
v.
United States,
470 U. S. 598, 607
(1985);
United States
v.
Goodwin,
457 U. S. 368, 380, n. 11
(1982);
Bordenkircher
v.
Hayes,
434 U. S. 357, 365
(1978). See also ABA Standards for Criminal Justice 3-3.8, 3-3.9 (2d ed. 1982).
Requiring a prosecutor to rebut a study that analyzes the past conduct of scores of prosecutors is quite different from requiring a prosecutor to rebut a contemporaneous challenge to his own acts. See
Batson
v.
Kentucky,
476 U. S. 79
(1986).
Although
Imbler
was decided in the context of damages actions under
42 U. S. C. § 1983
brought against prosecutors, the considerations that led the Court to hold that a prosecutor should not be required to explain his decisions apply in this case as well: “[I]f the prosecutor could be made to answer in court each time ... a person charged him with wrongdoing, his energy and attention would be diverted from the pressing duty of enforcing the criminal law.”
424 U. S., at 425
. Our refusal to require that the prosecutor provide an explanation for his decisions in this case is completely consistent with this Court’s longstanding precedents that hold that a prosecutor
*297
need not explain his decisions unless the criminal defendant presents a prima facie case of unconstitutional conduct with respect to his case. See,
e. g., Batson
v.
Kentucky, supra;
Wayte
v.
United States, supra.
In his dissent, Justice Blackmun misreads this statement. See
post,
at 348-349. We do not suggest that McCleskey’s conviction and sentencing by a jury bears on the prosecutor’s motivation. Rather, the fact that the United States Constitution and the laws of Georgia authorized the prosecutor to seek the death penalty under the circumstances of this case
is
a relevant factor to be weighed in determining whether the Baldus study demonstrates a constitutionally significant risk that this decision was motivated by racial considerations.
McCleskey relies on “historical evidence” to support his claim of purposeful discrimination by the State. This evidence focuses on Georgia laws in force during and just after the Civil War. Of course, the “historical background of the decision is one evidentiary source” for proof of intentional discrimination.
Arlington Heights
v.
Metropolitan Housing Dev. Corp.,
429 U. S., at 267
. But unless historical evidence is reasonably contemporaneous with the challenged decision, it has little probative value. Cf.
Hunter
v.
Underwood,
471 U. S. 222, 228-238
(1985) (relying on legislative history to demonstrate discriminatory motivation behind state statute). Although the history of racial discrimination in this country is undeniable, we cannot accept official actions taken long ago as evidence of current intent.
Justice Blackmun suggests that our “reliance on legitimate interests underlying the Georgia Legislature’s enactment of its capital punishment statute is . . . inappropriate [because] it has no relevance in a case dealing with a challenge to the Georgia capital sentencing system
as applied
in McCleskey’s case.”
Post,
at 349 (emphasis in original). As the dissent suggests, this evidence is not particularly probative when assessing the application of Georgia’s capital punishment system through the actions of prosecutors and juries, as we did in Part II-A,
supra.
But that is not the challenge that we are addressing here. As indicated above, the question we are addressing is whether the legislature maintains its capital punishment statute because of the racially disproportionate impact suggested by the Baldus study. McCleskey has introduced no evidence to support this claim. It is entirely appropriate to rely on the legislature’s legitimate reasons for enacting and maintaining a capital punishment statute to address a challenge to the
legislature’s
intent.
The Eighth Amendment applies to the States through the Due Process Clause of the Fourteenth Amendment.
Robinson
v.
California,
370 U. S. 660, 667
(1962).
Thirty-seven States now have capital punishment statutes that were enacted since our decision in
Furman.
Thirty-three of these States have imposed death sentences under the new statutes. NAACP Legal Defense and Educational Fund, Death Row, U. S. A. 1 (Oct. 1, 1986). A federal statute, amended in relevant part in 1974, authorizes the death penalty for aircraft piracy in which a death occurs. 49 U. S. C. App. § 1472(i)(1)(b).
We have noted that the Georgia statute generally follows the standards of the ALI Model Penal Code § 201.6 (Proposed Official Draft No. 13, 1961).
Gregg
v.
Georgia,
428 U. S., at 194, n. 44
.
Although the Court has recognized that jury sentencing in a capital case “can perform an important societal function-,”
Proffitt
v.
Florida,
428
*304
U. S. 242, 252
(1976) (joint opinion of Stewart, Powell, and Stevens, JJ.) (citing
Witherspoon
v.
Illinois,
391 U. S. 510, 519, n. 15
(1968)), it “has never suggested that jury sentencing [in a capital case] is constitutionally required.” 428 U. S., at 252. Under the Florida capital punishment system at issue in
Proffitt
,
the jury’s verdict is only advisory. The trial judge determines the final sentence. Unlike in Georgia, a Florida trial judge may impose the death penalty even when the jury recommends otherwise. In
Proffitt
,
we found that the Florida capital sentencing procedures adequately channeled the trial judge’s discretion so that the Florida system, like the Georgia system, on its face “satisfie[d] the constitutional deficiencies identified in
Furman.” Id.,
at 253
.
We have not yet decided whether the Constitution permits a mandatory death penalty in certain narrowly defined circumstances, such as when an inmate serving a life sentence without possibility of parole commits murder. See
Shuman
v.
Wolff,
791 F. 2d 788
(CA9), cert. granted
sub nom. Sumner
v.
Shuman,
479 U. S. 948
(1986).
This section is substantially identical to the current Georgia Code Ann. § 17-10-30(b)(7) (1982), which is reprinted in n. 3,
supra.
The Constitution is not offended by inconsistency in results based on the objective circumstances of the crime. Numerous legitimate factors may influence the outcome of a trial and a defendant’s ultimate sentence, even though they may be irrelevant to his actual guilt. If sufficient evidence to link a suspect to a crime cannot be found, he will not be charged. The capability of the responsible law enforcement agency can vary widely. Also, the strength of the available evidence remains a variable throughout the criminal justice process and may influence a prosecutor’s decision to offer a plea bargain or to go to trial. Witness availability, credibility, and memory also influence the results of prosecutions. Finally, sentencing in state courts is generally discretionary, so a defendant’s ultimate sentence necessarily will vary according to the judgment of the sentenc
*308
ing authority. The foregoing factors necessarily exist in varying degrees throughout our criminal justice system.
According to Professor Baldus:
“McCleskey’s case falls in [a] grey area where . . . you would find the greatest likelihood that some inappropriate consideration may have come to bear on the decision.
“In an analysis of this type, obviously one cannot say that we can say to a moral certainty what it was that influenced the decision. We can’t do that.” App. 45-46.
This Court has repeatedly stated that prosecutorial discretion cannot be exercised on the basis of race.
Wayte
v.
United States,
470 U. S., at 608
;
United States
v.
Batchelder,
442 U. S. 114
(1979);
Oyler
v.
Boles,
368 U. S. 448
(1962). Nor can a prosecutor exercise peremptory challenges on the basis of race.
Batson
v.
Kentucky,
476 U. S. 79
(1986);
Swain
v.
Alabama,
380 U. S. 202
(1966). More generally, this Court has condemned state efforts to exclude blacks from grand and petit juries.
Vasquez
v.
Hillery,
474 U. S. 254
(1986);
Alexander
v.
Louisiana,
405 U. S. 625, 628-629
(1972);
Whitus
v.
Georgia,
385 U. S., at 549-550
;
Norris v. Alabama,
294 U. S. 587, 589
(1935);
Neal
v.
Delaware,
103 U. S. 370, 394
(1881);
Strauder
v.
West Virginia,
100 U. S. 303, 308
(1880);
Ex parte Virginia,
100 U. S. 339
(1880).
Other protections apply to the trial and jury deliberation process. Widespread bias in the community can make a change of venue constitutionally required.
Irvin
v.
Dowd,
366 U. S. 717
(1961). The Constitution prohibits racially biased prosecutorial arguments.
Donnelly
v.
DeChristoforo,
416 U. S. 637, 643
(1974). If the circumstances of a particular case indicate a significant likelihood that racial bias may influence a jury, the Constitution requires questioning as to such bias.
Ristaino
v.
Ross,
424 U. S. 589, 596
(1976). Finally, in a capital sentencing hearing, a defendant convicted of an interracial murder is entitled to such questioning without regard to the circumstances of the particular case.
Turner
v.
Murray,
476 U. S. 28
(1986).
In advocating the adoption of the Constitution, Alexander Hamilton stated:
“The friends and adversaries of the plan of the convention, if they agree in nothing else, concur at least in the value they set upon the trial by jury; or if there is any difference between them, it consists in this: the former regard it as a valuable safeguard to liberty, the latter represent it as the very palladium of free government.” The Federalist No. 83, p. 519 (J. Gideon ed. 1818). '
In
Witherspoon
,
Justice Brennan joined the opinion of the Court written by Justice Stewart. The Court invalidated a statute that permitted a prosecutor to eliminate prospective jurors by challenging all who expressed qualms about the death penalty. The Court expressly recognized that the purpose of the “broad discretion” given to a sentencing jury is “to decide whether or not death is ‘the proper penalty’ in a given case,” noting that “a juror’s general views about capital punishment play an inevitable role in any such decision.” 391 U. S., at 519 (emphasis omitted). Thus, a sentencing jury must be composed of persons capable of expressing the “conscience of the community on the ultimate question of life or death.”
Ibid.
The Court referred specifically to the plurality opinion of Chief Justice Warren in
Trop
v.
Dulles,
356 U. S. 86
(1958), to the effect that it is the jury that must “maintain a link between contemporary community values and the penal system . . . .” 391 U. S., at 519, n. 15.
Justice Brennan’s condemnation of the results of the Georgia capital punishment system must be viewed against this background. As to community values and the constitutionality of capital punishment in general, we have previously noted, n. 23,
supra,
that the elected representatives of the people in 37 States and the Congress have enacted capital punishment statutes, most of which have been enacted or amended to conform generally to the
Gregg
standards, and that 33 States have imposed death sen
*311
tences thereunder. In the individual ease, a jury sentence reflects the conscience of the community as applied to the circumstances of a particular offender and offense. We reject Justice Beennan’s contention that this important standard for assessing the constitutionality of a death penalty should be abandoned.
In the guilt phase of a trial, the Double Jeopardy Clause bars re-prosecution after an acquittal, even if the acquittal is “ ‘based upon an egregiously erroneous foundation.’”
United States
v.
DiFrancesco,
449 U. S. 117, 129
(1980) (quoting
Fong Foo
v.
United States,
369 U. S. 141, 143
(1962)). See Powell, Jury Trial of Crimes,
23 Wash. & Lee L. Rev. 1
, 7-8 (1966) (Despite the apparent injustice of such an acquittal, “[t]he founding
*312
fathers, in light of history, decided that the balance here should be struck in favor of the individual”).
In the penalty hearing, Georgia law provides that “unless the jury . . . recommends the death sentence in its verdict, the court shall not sentence the defendant to death.” Georgia Code Ann. § 17-10-31 (1982). In
Bullington
v.
Missouri,
451 U. S. 430
(1981), this Court held that the Double Jeopardy Clause of the Constitution prohibits a State from asking for a sentence of death at a second trial when the jury at the first trial recommended a lesser sentence.
In this ease, for example, McCleskey declined to enter a guilty plea. According to his trial attorney: “[T]he Prosecutor was indicating that we might be able to work out a life sentence if he were willing to enter a plea. But we never reached any concrete stage on that because Mr. MeCleskey’s attitude was that he didn’t want to enter a plea. So it never got any further than just talking about it.” Tr. in No. 4909, p. 56 (Jan. 30, 1981).
Congress has acknowledged the existence of such discrepancies in criminal sentences, and in 1984 created the United States Sentencing Commission to develop sentencing guidelines. The objective.of the guidelines “is to avoid
unwarranted
sentencing disparities among defendants with similar records who have been found guilty of similar criminal conduct, while maintaining sufficient flexibility to permit individualized sentencing when warranted by mitigating or aggravating factors not taken into account in the guidelines.”
52 Fed. Reg. 3920
(1987) (emphasis added). No one contends that all sentencing disparities can be eliminated. The
*313
guidelines, like the safeguards in the Gregg-type statute, further an essential need of the Anglo-American criminal justice system — to balance the desirability of a high degree of uniformity against the necessity for the exercise of discretion.
The Baldus study in fact confirms that the Georgia system results in a reasonable level of proportionality among the class of murderers eligible for the death penalty. As Professor Baldus confirmed, the system sorts out cases where the sentence of death is highly likely and highly unlikely, leaving a midrange of eases where the imposition of the death penalty in any particular case is less predictable. App. 35-36. See n. 5,
supra.
Justice BRENNAN’s eloquent dissent of course reflects his often repeated opposition to the death sentence. His views, that also are shared by Justice MARSHALL, are principled and entitled to respect. Neverthe
*314
less, since
Gregg
was decided in 1976, seven Members of this Court consistently have upheld sentences of death under
Gregg-type
statutes providing for meticulous review of each sentence in both state and federal courts. The ultimate thrust of Justice Brennan’s dissent is that
Gregg
and its progeny should be overruled. He does not, however, expressly call for the overruling of any prior decision. Rather, relying on the Baldus study, Justice Brennan, joined by Justices Marshall, Blackmun, and Stevens, questions the very heart of our criminal justice system: the traditional discretion that prosecutors and juries necessarily must have.
We have held that discretion in a capital punishment system is necessary to satisfy the Constitution.
Woodson
v.
North Carolina,
428 U. S. 280
(1976). See
supra,
at 303-306. Yet, the dissent now claims that the “discretion afforded prosecutors and jurors in the Georgia capital sentencing system” violates the Constitution by creating “opportunities for racial considerations to influence criminal proceedings.”
Post,
at 333. The dissent contends that in Georgia “[n]o guidelines govern prosecutorial decisions . . . and [that] Georgia provides juries with no list of aggravating and mitigating factors, nor any standard for balancing them against one another.”
Ibid.
Prosecutorial decisions necessarily involve both judgmental and factual decisions that vary from case to ease. See ABA Standards for Criminal Justice 3-3.8, 3-3.9 (2d ed. 1982). Thus, it is difficult to imagine guidelines that would produce the predictability sought by the dissent without sacrificing the discretion essential to a humane and fair system of criminal justice. Indeed, the dissent suggests no such guidelines for pros-ecutorial discretion.
The reference to the failure to provide juries with the list of aggravating and mitigating factors is curious. The aggravating circumstances are set forth in detail in the Georgia statute. See n. 3,
supra.
The jury is not provided with a list of aggravating circumstances because not all of them are relevant to any particular crime. Instead, the prosecutor must choose the relevant circumstances and the State must prove to the jury that at least one exists beyond a reasonable doubt before the jury can even consider imposing the death sentence. It would be improper and often prejudicial to allow jurors to speculate as to aggravating circumstances wholly without support in the evidence.
The dissent’s argument that a list of mitigating factors is required is particularly anomalous. We have held that the Constitution requires that juries be allowed to consider “any relevant mitigating factor,” even if it is not
*315
included in a statutory list.
Eddings
v.
Oklahoma,
455 U. S., at 112
. See
Lockett
v.
Ohio,
438 U. S. 586
(1978). The dissent does not attempt to harmonize its criticism with this constitutional principle. The dissent also does not suggest any standard, much less a workable one, for balancing aggravating and mitigating factors. If capital defendants are to be treated as “uniquely individual human beings,”
Woodson
v.
North Carolina, supra,
at 304, then discretion to evaluate and weigh the circumstances relevant to the particular defendant and the crime he committed is essential.
The dissent repeatedly emphasizes the need for “a uniquely high degree of rationality in imposing the death penalty.”
Post,
at 335. Again, no suggestion is made as to how greater “rationality” could be achieved under any type of statute that authorizes capital punishment. The Gregg-type statute imposes unprecedented safeguards in the special context of capital punishment. These include: (i) a bifurcated sentencing proceeding; (ii) the threshold requirement of one or more aggravating circumstances; and (iii) mandatory State Supreme Court review. All of these are administered pursuant to this Court’s decisions interpreting the limits of the Eighth Amendment on the imposition of the death penalty, and all are subject to ultimate review by this Court. These ensure a degree of care in the imposition of the sentence of death that can be described only as unique. Given these safeguards already inherent in the imposition and review of capital sentences, the dissent’s call for greater rationality is no less than a claim that a capital punishment system cannot be administered in accord with the Constitution. As we reiterate,
infra,
the requirement of heightened rationality in the imposition of capital punishment does not “plae[e] totally unrealistic conditions on its use.”
Gregg
v.
Georgia,
428 U. S., at 199, n. 50
.
Studies already exist that allegedly demonstrate a racial disparity in the length of prison sentences. See,
e. g.,
Spohn, Gruhl, & Wélch, The Effect of Race on Sentencing: A Reexamination of an Unsettled Question, 16 Law & Soc. Rev. 71 (1981-1982); Unnever, Frazier, & Henretta, Race Differences in Criminal Sentencing, 21 Sociological Q. 197 (1980).
In
Regents of the University of California
v.
Bakke,
438 U. S. 265, 295
(1978) (opinion of Powell, J.), we recognized that the national “majority” “is composed of various minority groups, most of which can lay claim to a history of prior discrimination at the hands of the State and private individuals.” See
id.,
at 292 (citing
Strauder
v.
West Virginia,
100 U. S., at 308
(Celtic Irishmen) (dictum);
Yick Wo
v.
Hopkins,
118 U. S. 356
(1886) (Chinese);
Truax
v.
Raich,
239 U. S. 33, 36, 41-42
(1915) (Austrian resident aliens);
Korematsu
v.
United States,
323 U. S. 214, 216
(1944) (Japanese);
Hernandez
v.
Texas,
347 U. S. 475
(1954) (Mexican-Americans)). See also Uniform Guidelines on Employee Selection Procedures (1978),
29 CFR § 1607.4
(B) (1986) (employer must keep records as to the “following races and ethnic groups: Blacks, American Indians (including Alaskan Natives), Asians (including Pacific Islanders), Hispanics (including persons of Mexican, Puerto Rican, Cuban, Central or South American, or other Spanish origin or culture regardless of race), and whites (Caucasians) other than Hispanics”); U. S. Bureau of the Census, 1980 Census of the Population, Vol. 1, ch. B (PC80-1-B), reprinted in 1986 Statistical Abstract of the United States 29 (dividing United States population by “race and Spanish origin” into the following groups: White, Black, American Indian, Chinese, Filipino, Japanese, Korean, Vietnamese, Spanish origin, and all other races); U. S. Bureau of the Census, 1980 Census of the Population, Supplementary Report, series PC80-S1-10, reprinted in 1986 Statistical Abstract of the United States 34 (listing 44 ancestry groups and noting that many individuals reported themselves to belong to multiple ancestry groups).
We also have recognized that the ethnic composition of the Nation is ever shifting.
Crawford
v.
Board of Ed. of Los Angeles,
458 U. S. 527
(1982), illustrates demographic facts that we increasingly find in our country, namely, that populations change in composition, and may do so in relatively short timespans. We noted: “In 1968 when the case went to trial, the [Los Angeles] District was 53.6% white, 22.6% black, 20% Hispanic, and 3.8% Asian and other. By October 1980, the demographic composition had altered radically: 23.7% white, 23.3% black, 45.3% Hispanic, and 7.7% Asian andother.” M,at530, n. 1. Increasingly whites are becoming a minority in many of the larger American cities. There appears to be no reason why a white defendant in such a city could not make a claim similar to McCleskey’s if racial disparities in sentencing arguably are shown by a statistical study.
Finally, in our heterogeneous society the lower courts have found the boundaries of race and ethnicity increasingly difficult to determine. See
*317
Shaare Tefila Congregation
v.
Cobb,
785 F. 2d 523
(CA4), cert. granted,
479 U. S. 812
(1986), and
Al-Khazraji
v.
Saint Francis College,
784 F. 2d 505
(CA3), cert. granted,
479 U. S. 812
(1986) (argued Feb. 25, 1987) (presenting the questions whether Jews and Arabs, respectively, are “races” covered by
42 U. S. C. §§ 1981
and 1982).
See Chamblin, The Effect of Sex on the Imposition of the Death Penalty (speech given at a symposium of the American Psychological Association, entitled “Extra-legal Attributes Affecting Death Penalty Sentencing,” New York City, Sept., 1979); Steffensmeier, Effects of Judge’s and Defendant’s Sex on the Sentencing of Offenders, 14 Psychology, Journal of Human Behavior, 3 (Aug. 1977).
See Johnson, Black Innocence and the White Jury,
83 Mich. L. Rev. 1611
, 1625-1640, and n. 115 (1985) (citing Cohen & Peterson, Bias in the Courtroom: Race and Sex Effects of Attorneys on Juror Verdicts, 9 Social Behavior & Personality 81 (1981)); Hodgson & Pryor, Sex Discrimination in the Courtroom: Attorney’s Gender and Credibility, 55 Psychological Rep. 483 (1984).
See Steffensmeier,
supra,
at 7.
See Kerr, Bull, MacCoun, & Rathborn, Effects of victim attractiveness, care and disfigurement on the judgements of American and British mock jurors, 24 Brit. J. Social Psych. 47 (1985); Johnson,
supra,
at 1638, n. 128 (citing Shoemaker, South, & Lowe, Facial Stereotypes of Deviants and Judgments of Guilt or Innocence, 51 Social Forces 427 (1973)).
Some studies indicate that physically attractive defendants receive greater leniency in sentencing than unattractive defendants, and that offenders whose victims are physically attractive receive harsher sentences than defendants with less attractive victims. Smith & Hed, Effects of Offenders’ Age and Attractiveness on Sentencing by Mock Juries, 44 Psy
*318
chological Rep. 691 (1979); Kerr, Beautiful and Blameless: Effects of Victim Attractiveness and Responsibility on Mock Jurors’ Verdicts, 4 Personality and Social Psych. Bull. 479 (1978). But see Baumeister & Darley, Reducing the Biasing Effect of Perpetrator Attractiveness in Jury Simulation, 8 Personality and Social Psych. Bull. 286 (1982); Schwibbe & Schwibbe, Judgment and Treatment of People of Varied Attractiveness, 48 Psychological Rep. 11 (1981); Weiten, The Attraction-Leniency Effect in Jury Research: An Examination of External Validity, 10 J. Applied Social Psych. 340 (1980).
Justice Stevens, who would not overrule
Gregg
,
suggests in his dissent that the infirmities alleged by McCleskey could be remedied by narrowing the class of death-eligible defendants to categories identified by the Baldus study where “prosecutors consistently seek, and juries consistently impose, the death penalty without regard to the race of the victim or the race of the offender.”
Post,
at 367. This proposed solution is unconvincing. First, “consistently” is a relative term, and narrowing the category of death-eligible defendants would simply shift the borderline between those defendants who received the death penalty and those who did not. A borderline area would continue to exist and vary in its boundaries. Moreover, because the discrepancy between borderline cases would be difficult to explain, the system would likely remain open to challenge on the basis that the lack of explanation rendered the sentencing decisions unconstitutionally arbitrary.
Second, even assuming that a category with theoretically consistent results could be identified, it is difficult to imagine how Justice Stevens’ proposal would or could operate on a ease-by-case basis. Whenever a victim is white and the defendant is a member of a different race, what steps would a prosecutor be required to take-in addition to weighing the customary prosecutorial considerations — before concluding in the particular case that he lawfully could prosecute? In the absence of a current, Baldus-type study focused particularly on the community in which the crime was committed, where would he find a standard? Would the prosecutor have to review the prior decisions of community prosecutors and determine the types of eases in which juries in his jurisdiction “consistently” had imposed the death penalty when the victim was white and the defendant was of a different race? And must he rely solely on statistics? Even if such a study were feasible, would it be unlawful for the prosecutor, in making his
*319
final decision in a particular case, to consider the evidence of guilt and the presence of aggravating and mitigating factors? However conscientiously a prosecutor might attempt to identify death-eligible defendants under the dissent’s suggestion, it would be a wholly speculative task at best, likely to result in less rather than more fairness and consistency in the imposition of the death penalty.

=== Dissent ===

Justice Brennan,
with whom Justice Marshall joins, and with whom Justice Blackmun and Justice Stevens join in all but Part I, dissenting.
I
Adhering to my view that the death penalty is in all circumstances cruel and unusual punishment forbidden by the Eighth and Fourteenth Amendments, I would vacate the decision below insofar as it left undisturbed the death sentence imposed in this case.
Gregg
v.
Georgia,
428 U. S. 153, 227
(1976) (Brennan, J., dissenting). The Court observes that “[t]he Gregg-type statute imposes unprecedented safeguards in the special context of capital punishment,” which “ensure a degree of care in the imposition of the death penalty that can be described only as unique.”
Ante,
at 315, n. 37. Notwithstanding these efforts, murder defendants in Georgia with white victims are more than four times as likely to receive the death sentence as are defendants with black victims. Petitioner’s Exhibit DB 82. Nothing could convey more powerfully the intractable reality of the death penalty: “that the effort to eliminate arbitrariness in the infliction of that ultimate sanction is so plainly doomed to failure that it — and the death penalty — must be abandoned altogether.”
Godfrey
v.
Georgia,
446 U. S. 420, 442
(1980) (Marshall, J., concurring in judgment).
Even if I did not hold this position, however, I would reverse the Court of Appeals, for petitioner McCleskey has clearly demonstrated that his death sentence was imposed in violation of the Eighth and Fourteenth Amendments. While I join Parts I through IV-A of Justice Blackmun’s dissenting opinion discussing petitioner’s Fourteenth Amendment claim, I write separately to emphasize how conclusively
*321
McCleskey has also demonstrated precisely the type of risk of irrationality in sentencing that we have consistently condemned in our Eighth Amendment jurisprudence.
At some point in this case, Warren McCleskey doubtless asked his lawyer whether a jury was likely to sentence him to die. A candid reply to this question would have been disturbing. First, counsel would have to tell McCleskey that few of the details of the crime or of McCleskey’s past criminal conduct were more important than the fact that his victim was white. Petitioner’s Supplemental Exhibits (Supp. Exh.) 50. Furthermore, counsel would feel bound to tell McCleskey that defendants charged with killing white victims in Georgia are 4.3 times as likely to be sentenced to death as defendants charged with killing blacks. Petitioner’s Exhibit DB 82. In addition, frankness would compel the disclosure that it was more likely than not that the race of McCleskey’s victim would determine whether he received a death sentence: 6 of every 11 defendants convicted of killing a white person would not have received the death penalty if their victims had been black, Supp. Exh. 51, while, among defendants with aggravating and mitigating factors comparable to McCleskey’s, 20 of every 34 would not have been sentenced to die if their victims had been black.
Id.,
at 54. Finally, the assessment would not be complete without the information that cases involving black defendants and white victims are more likely to result in a death sentence than cases featuring any other racial combination of defendant and victim.
Ibid.
The story could be told in a variety of ways, but McCleskey could not fail to grasp its essential narrative line: there was a significant chance that race would play a prominent role in determining if he lived or died.
The Court today holds that Warren McCleskey’s sentence was constitutionally imposed. It finds no fault in a system in which lawyers must tell their clients that race casts a
*322
large shadow on the capital sentencing process. The Court arrives at this conclusion by stating that the Baldus study cannot
“prove
that race enters into any capital sentencing decisions or that race was a factor in McCleskey’s particular case.”
Ante,
at 308 (emphasis in original). Since, according to Professor Baldus, we cannot say “to a moral certainty” that race influenced a decision,
ante,
at 308, n. 29, we can identify only “a likelihood that a particular factor entered into some decisions,”
ante,
at 308, and “a discrepancy that appears to correlate with race.”
Ante,
at 312. This “likelihood” and “discrepancy,” holds the Court, is insufficient to establish a constitutional violation. The Court reaches this conclusion by placing four factors on the scales opposite Mc-Cleskey’s evidence: the desire to encourage sentencing discretion, the existence of “statutory safeguards” in the Georgia scheme, the fear of encouraging widespread challenges to other sentencing decisions, and the limits of the judicial role. The Court’s evaluation of the significance of petitioner’s evidence is fundamentally at odds with our consistent concern for rationality in capital sentencing, and the considerations that the majority invokes to discount that evidence cannot justify ignoring its force.
Ill
A
It is important to emphasize at the outset that the Court’s observation that McCleskey cannot prove the influence of race on any particular sentencing decision is irrelevant in evaluating his Eighth Amendment claim. Since
Furman
v.
Georgia,
408 U. S. 238
(1972), the Court has been concerned with the
risk
of the imposition of an arbitrary sentence, rather than the proven fact of one.
Furman
held that the death penalty “may not be imposed under sentencing procedures that create a substantial risk that the punishment will be inflicted in an arbitrary and capricious manner.”
Godfrey
v.
Georgia, supra,
at 427. As Justice O’Connor observed
*323
in
Caldwell
v.
Mississippi,
472 U. S. 320, 343
(1985), a death sentence must be struck down when the circumstances under which it has been imposed “creat[e] an unacceptable
risk
that ‘the death penalty [may have been] meted out arbitrarily or capriciously’ or through ‘whim or mistake’ ” (emphasis added) (quoting
California
v.
Ramos,
463 U. S. 992, 999
(1983)). This emphasis on risk acknowledges the difficulty of divining the jury’s motivation in an individual case. In addition, it reflects the fact that concern for arbitrariness focuses on the rationality of the system as a whole, and that a system that features a significant probability that sentencing decisions are influenced by impermissible considerations cannot be regarded as rational.
1
As we said in
Gregg
v.
Georgia,
428 U. S., at 200
, “the petitioner looks to the sentencing system as a whole (as the Court did in
Furman
and we do today)”: a constitutional violation is established if a plaintiff demonstrates a
“pattern
of arbitrary and capricious sentencing.”
Id.,
at 195, n. 46 (emphasis added) (joint opinion of Stewart, Powell, and Stevens, JJ.).
As a result, our inquiry under the Eighth Amendment has not been directed to the validity of the individual sentences before us. In
Godfrey
,
for instance, the Court struck down the petitioner’s sentence because the vagueness of the statutory definition of heinous crimes created a
risk
that prejudice
*324
or other impermissible influences
might have infected
the sentencing decision. In vacating the sentence, we did not ask whether it was likely that Godfrey’s own sentence reflected the operation of irrational considerations. Nor did we demand a demonstration that such considerations had actually entered into other sentencing decisions involving heinous crimes. Similarly, in
Roberts
v.
Louisiana,
428 U. S. 825
(1976), and
Woodson
v.
North Carolina,
428 U. S. 280
(1976), we struck down death sentences in part because mandatory imposition of the death penalty created the
risk
that a jury
might
rely on arbitrary considerations in deciding which persons should be convicted of capital crimes. Such a risk would arise, we said, because of the likelihood that jurors reluctant to impose capital punishment on a particular defendant would refuse to return a conviction, so that the effect of mandatory sentencing would be to recreate the unbounded sentencing discretion condemned in
Furman. Roberts, supra,
at 334-335 (plurality opinion);
Woodson, supra,
at 303
(plurality opinion). We did not ask whether the death sentences in the cases before us could have reflected the jury’s rational consideration and rejection of mitigating factors. Nor did we require proof that juries had actually acted irrationally in other cases.
Defendants challenging their death sentences thus never have had to prove that impermissible considerations have actually infected sentencing decisions. We have required instead that they establish that the system under which they were sentenced posed a significant risk of such an occurrence. McCleskey’s claim does differ, however, in one respect from these earlier cases: it is the first to base a challenge not on speculation about how a system
might
operate, but on empirical documentation of how it
does
operate.
The Court assumes the statistical validity of the Baldus study, and acknowledges that McCleskey has demonstrated a risk that racial prejudice plays a role in capital sentencing in Georgia,
ante,
at 291, n. 7. Nonetheless, it finds the probability of prejudice insufficient to create constitutional con
*325
cern.
Ante,
at 313. Close analysis of the Baldus study, however, in light of both statistical principles and human experience, reveals that the risk that race influenced Mc-Cleskey’s sentence is intolerable by any imaginable standard.
B
The Baldus study indicates that, after taking into account some 230 nonracial factors that might legitimately influence a sentencer, the jury
more likely than not
would have spared McCleskey’s life had his victim been black. The study distinguishes between those cases in which (1) the jury exercises virtually no discretion because the strength or weakness of aggravating factors usually suggests that only one outcome is appropriate;
2
and (2) cases reflecting an “intermediate” level of aggravation, in which the jury has considerable discretion in choosing a sentence.
3
McCleskey’s case falls into the intermediate range. In such cases, death is imposed in 34% of white-victim crimes and 14% of black-victim crimes, a difference of 139% in the rate of imposition of the death penalty. Supp. Exh. 54. In other words, just under 59% — almost 6 in 10 — defendants comparable to McCleskey would not have received the death penalty if their victims had been black.
4
*326
Furthermore, even examination of the sentencing system as a whole, factoring in those cases in which the jury exercises little discretion, indicates the influence of race on capital sentencing. For the Georgia system as a whole, race accounts for a six percentage point difference in the rate at which capital punishment is imposed. Since death is imposed in 11% of all white-victim cases, the rate in comparably aggravated black-victim cases is 5%. The rate of capital sentencing in a white-victim case is thus 120% greater than the rate in a black-victim case. Put another way, over half— 55% — of defendants in white-victim crimes in Georgia would not have been sentenced to die if their victims had been black. Of the more than 200 variables potentially relevant to a sentencing decision, race of the victim is a powerful explanation for variation in death sentence rates — as powerful as nonracial aggravating factors such as a prior murder conviction or acting as the principal planner of the homicide.
5
These adjusted figures are only the most conservative indication of the risk that race will influence the death sentences of defendants in Georgia. Data unadjusted for the mitigating or aggravating effect of other factors show an even more pronounced disparity by race. The capital sentencing rate for all white-victim cases was almost
11 times
greater than
*327
the rate for black-victim cases. Supp. Exh. 47. Furthermore, blacks who kill whites are sentenced to death at nearly
22 times
the rate of blacks who kill blacks, and more than 7
times
the rate of whites who kill blacks.
Ibid.
In addition, prosecutors seek the death penalty for 70% of black defendants with white victims, but for only 15% of black defendants with black victims, and only 19% of white defendants with black victims.
Id.,
at 56. Since our decision upholding the Georgia capital sentencing system in
Gregg
,
the State has executed seven persons. All of the seven were convicted of killing whites, and six of the seven executed were black.
6
Such execution figures are especially striking in light of the fact that, during the period encompassed by the Baldus study, only 9.2% of Georgia homicides involved black defendants and white victims, while 60.7% involved black victims.
McCleskey’s statistics have particular force because most of them are the product of sophisticated multiple-regression analysis. Such analysis is designed precisely to identify patterns in the aggregate, even though we may not be able to reconstitute with certainty any individual decision that goes to make up that pattern. Multiple-regression analysis is particularly well suited to identify the influence of impermissible considerations in sentencing, since it is able to control for permissible factors that may explain an apparent arbitrary pattern.
7
While the decisionmaking process of a body such as a jury may be complex, the Baldus study provides a massive compilation of the details that are most relevant to that decision. As we held in the context of Title VII of the Civil Rights Act of 1964 last Term in
Bazemore
v.
Friday,
478 U. S. 385
(1986), a multiple-regression analysis need not include every conceivable variable to establish a party’s case, as long as it includes those variables that account for the
*328
major factors that are likely to influence decisions. In this case, Professor Baldus in fact conducted additional regression analyses in response to criticisms and suggestions by the District Court, all of which confirmed, and some of which even strengthened, the study’s ori