with whom Justice Marshall and Justice Stevens join, and with whom Justice Brennan joins in all but Part IV-B, dissenting.
The Court today sanctions the execution of a man despite his presentation of evidence that establishes a constitutionally intolerable level of racially based discrimination leading to the imposition of his death sentence. I am disappointed with the Court’s action not only because of its denial of constitutional guarantees to petitioner McCleskey individually, but also because of its departure from what seems to me to be well-developed constitutional jurisprudence.
Justice Brennan has thoroughly demonstrated, ante, that, if one assumes that the statistical evidence presented by petitioner McCleskey is valid, as we must in light of the Court of Appeals’ assumption,1 there exists in the Georgia capital sentencing scheme a risk of racially based discrimination that is so acute that it violates the Eighth Amendment. His analysis of McCleskey’s case in terms of the Eighth Amendment is consistent with this Court’s recognition that because.capital cases involve the State’s imposition of a punishment that is unique both in kind and degree, the decision in such cases must reflect a heightened degree of reliability under the Amendment’s prohibition of the infliction of cruel and unusual punishments. See Woodson v. North Carolina, 428 U. S. 280, 305 (1976) (plurality opinion). I therefore join Parts II through V of Justice Brennan’s dissenting opinion.
*346Yet McCleskey’s case raises concerns that are central not only to the principles underlying the Eighth Amendment, but also to the principles underlying the Fourteenth Amendment. Analysis of his case in terms of the Fourteenth Amendment is consistent with this Court’s recognition that racial discrimination is fundamentally at odds with our constitutional guarantee of equal protection. The protections afforded by the Fourteenth Amendment are not left at the courtroom door. Hill v. Texas, 316 U. S. 400, 406 (1942). Nor is equal protection denied to persons convicted of crimes. Lee v. Washington, 390 U. S. 333 (1968) (per curiam). The Court in the past has found that racial discrimination within the criminal justice system is particularly abhorrent: “Discrimination on the basis of race, odious in all aspects, is especially pernicious in the administration of justice.” Rose v. Mitchell, 443 U. S. 546, 555 (1979). Disparate enforcement of criminal sanctions “destroys the appearance of justice and thereby casts doubt on the integrity of the judicial process.” Id., at 555-556. And only last Term Justice Powell, writing for the Court, noted: “Discrimination within the judicial system is most pernicious because it is ‘a stimulant to that race prejudice which is an impediment to securing to [black citizens] that equal justice which the law aims to secure to all others.’ ” Batson v. Kentucky, 476 U. S. 79, 87-88 (1986), quoting Strauder v. West Virginia, 100 U. S. 303, 308 (1880).
Moreover, the legislative history of the Fourteenth Amendment reminds us that discriminatory enforcement of States’ criminal laws was a matter of great concern for the drafters. In the introductory remarks to its Report to Congress, the Joint Committee on Reconstruction, which reported out the Joint Resolution proposing the Fourteenth Amendment, specifically noted: “This deep-seated prejudice against color . . . leads to acts of cruelty, oppression, and murder, which the local authorities are at no pains to prevent or punish.” H. R. Joint Comm. Rep. No. 30, 39th Cong., 1st Sess., p. XVII (1866). Witnesses who testified before *347the Committee presented accounts of criminal acts of violence against black persons that were not prosecuted despite evidence as to the identity of the perpetrators.2
A
The Court today seems to give a new meaning to our recognition that death is different. Rather than requiring *348“a correspondingly greater degree of scrutiny of the capital sentencing determination,” California v. Ramos, 463 U. S. 992, 998-999 (1983), the Court relies on the very fact that this is a case involving capital punishment to apply a lesser standard of scrutiny under the Equal Protection Clause. The Court concludes that “legitimate” explanations outweigh McCleskey’s claim that his death sentence reflected a constitutionally impermissible risk of racial discrimination. The Court explains that McCleskey’s evidence is too v/eak to require 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.” Ante, at 297. The Court states that it will not infer a discriminatory purpose on the part of the state legislature because “there were legitimate reasons for the Georgia Legislature to adopt and maintain capital punishment.” Ante, at 298-299.
The Court’s assertion that the fact of McCleskey’s conviction undermines his constitutional claim is inconsistent with a long and unbroken line of this Court’s case law. The Court on numerous occasions during the past century has recognized that an otherwise legitimate basis for a conviction does not outweigh an equal protection violation. In cases where racial discrimination in the administration of the criminal justice system is established, it has held that setting aside the conviction is the appropriate remedy. See, e. g., Rose v. Mitchell, 443 U. S., at 559; Whitus v. Georgia, 385 U. S. 545, 549-550 (1967); Strauder v. West Virginia, 100 U. S. 303 (1880). The Court recently reaffirmed the propriety of invalidating a conviction in order to vindicate federal constitutional rights. Vasquez v. Hillery, 474 U. S. 254 (1986). Invalidation of a criminal conviction on federal constitutional grounds does not necessarily preclude retrial and resentencing of the defendant by the State. Hill v. Texas, 316 U. S., at 406. The Court has maintained a per se reversal *349rule rejecting application of harmless-error analysis in cases involving racial discrimination that “strikes at the fundamental values of our judicial system and our society as a whole.” Rose v. Mitchell, 448 U. S., at 556. We have noted that a conviction “in no way suggests that the discrimination did not impermissibly infect” earlier phases of the criminal prosecution “and, consequently, the nature or very existence of the proceedings to come.” Vasquez v. Hillery, 474 U. S., at 263. Hence, McCleskey’s conviction and the imposition of his death sentence by the jury do not suggest that discrimination did not impermissibly infect the earlier steps in the prosecution of his case, such as the prosecutor’s decision to seek the death penalty.
The Court’s reliance on legitimate interests underlying the Georgia Legislature’s enactment of its capital punishment statute is likewise inappropriate. Although that reasoning may be relevant in a case involving a facial challenge to the constitutionality of a statute, it has no relevance in a case dealing with a challenge to the Georgia capital sentencing system as applied in McCleskey’s case. In Batson v. Kentucky, supra, we rejected such reasoning: “The Constitution requires . . . that we look beyond the face of the statute . . . and also consider challenged selection practices to afford ‘protection against action of the State through its administrative officers in effecting the prohibited discrimination.’” 476 U. S., at 88, quoting Norris v. Alabama, 294 U. S. 587, 589 (1935).
B
In analyzing an equal protection claim, a court must first determine the nature of the claim and the responsibilities of the state actors involved to determine what showing is required for the establishment of a prima facie case. Castaneda v. Partida, 430 U. S. 482, 493-494 (1977). The Court correctly points out: “In its broadest form, McCles-key’s claim of discrimination extends to every actor in the Georgia’capital sentencing process, from the prosecutor who *350sought 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.” Ante, at 292. Having recognized the complexity of McCleskey’s claim, however, the Court proceeds to ignore a significant element of that claim. The Court treats the case as if it is limited to challenges to the actions of two specific decisionmaking bodies — the petit jury and the state legislature. Ante, at 294-295, 297-298. This self-imposed restriction enables the Court to distinguish this case from the venire-selection cases and cases under Title VII of the Civil Rights Act of 1964 in which it long has accepted statistical evidence and has provided an easily applicable framework for review. See e. g., Castaneda v. Partida, supra; Bazemore v. Friday, 478 U. S. 385 (1986) (Brennan, J., joined by all other Members of the Court, concurring in part). Considering McCleskey’s claim in its entirety, however, reveals that the claim fits easily within that same framework. A significant aspect of 'his claim is that racial factors impermissibly affected numerous steps in the Georgia capital sentencing scheme between his indictment and the jury’s vote to sentence him to death. The primary decisionmaker at each of the intervening steps of the process is the prosecutor, the quintessential state actor in a criminal proceeding.3 The District Court expressly stated *351that there were “two levels of the system that matter to [MeCleskey], the decision to seek the death penalty and the decision to impose the death penalty.” 580 F. Supp. 338, 379-380 (ND Ga. 1984). I agree with this statement of McCleskey’s case. Hence, my analysis in this dissenting opinion takes into account the role of the prosecutor in the Georgia capital sentencing system. I certainly do not address all the alternative methods of proof in the Baldus study. Nor do I review each step in the process which MeCleskey challenges. I concentrate on the decisions within the prosecutor’s office through which the State decided to seek the death penalty and, in particular, the point at which the State proceeded to the penalty phase after conviction. This is a step at which the evidence of the effect of the racial factors was especially strong, see Supplemental Exhibits (Supp. Exh.) 56, 57; Transcript of Federal Habeas Corpus Hearing in No. C81-2434A (Tr.) 894-926, but is ignored by the Court.
H
>
A criminal defendant alleging an equal protection violation must prove the existence of purposeful discrimination. Washington v. Davis, 426 U. S. 229, 239-240 (1976); Whitus v. Georgia, 385 U. S., at 550. He may establish a prima facie case4 of purposeful discrimination “by showing that the *352totality of the relevant facts gives rise to an inference of discriminatory purpose.” Batson v. Kentucky, 476 U. S., at 94.5 Once the defendant establishes a prima facie case, the burden shifts to the prosecution to rebut that case. “The State cannot meet this burden on mere general assertions that its officials did not discriminate or that they properly performed their official duties.” Ibid. The State must demonstrate that the challenged effect was due to “ ‘permissible racially neutral selection criteria.’” Ibid., quoting Alexander v. Louisiana, 405 U. S. 625, 632 (1972).
Under Batson v. Kentucky and the framework established in Castaneda v. Partida, McCleskey must meet a three-factor standard. First, he must establish that he is a member of a group “that is a recognizable, distinct class, singled out for different treatment.” 430 U. S., at 494. Second, he must make a showing of a substantial degree of differential treatment.6 Third, he must establish that the allegedly *353discriminatory procedure is susceptible to abuse or is not racially neutral. Ibid.
B
There can be no dispute that McCleskey has made the requisite showing under the first prong of the standard. The Baldus study demonstrates that black persons are a distinct group that are singled out for different treatment in the Georgia capital sentencing system. The Court acknowledges, as it must, that the raw statistics included in the Bal-dus study and presented by petitioner indicate that it is much less likely that a death sentence will result from a murder of a black person than from a murder of a white person. Ante, at 286. White-victim cases are nearly 11 times more likely to yield a death sentence than are black-victim cases. Supp. Exh. 46. The raw figures also indicate that even within the group of defendants who are convicted of killing white persons and are thereby more likely to receive a death sentence, black defendants are more likely than white defendants to be sentenced to death. Supp. Exh. 47.
With respect to the second prong, McCleskey must prove that there is a substantial likelihood that his death sentence is due to racial factors. See Hunter v. Underwood, 471 U. S. 222, 228 (1985). The Court of Appeals assumed the validity of the Baldus study and found that it “showed that systemic 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.” 753 F. 2d 877, 895 (CA11 1985). *354The question remaining therefore is at what point does that disparity become constitutionally unacceptable. See Turner v. Murray, 476 U. S. 28, 36, n. 8 (1986) (plurality opinion). Recognizing that additional factors can enter into the deci-sionmaking process that yields a death sentence, the authors of the Baldus study collected data concerning the presence of other relevant factors in homicide cases in Georgia during the time period relevant to McCleskey’s case. They then analyzed the data in a manner that would permit them to ascertain the independent effect of the racial factors.7
McCleskey demonstrated the degree to which his death sentence was affected by racial factors by introducing multiple-*355regression analyses that explain how much of the statistical distribution of the cases analyzed is attributable to the racial factors. McCleskey established that because he was charged with killing a white person he was 4.3 times as likely to be sentenced to death as he would have been had he been charged with killing a black person. Petitioner’s Exhibit DB 82. McCleskey also demonstrated that it was more likely than not that the fact that the victim he was charged with killing was white determined that he received a sentence of death — 20 out of every 34 defendants in McCleskey’s mid-range category would not have been sentenced to be executed if their victims had been black. Supp. Exh. 54.8 The most persuasive evidence of the constitutionally significant effect of racial factors in the Georgia capital sentencing system is McCleskey’s proof that the race of the victim is more important in explaining the imposition of a death sentence than is the factor whether the defendant was a prime mover in the homicide. Petitioner’s Exhibit DB 82.9 Similarly, the race-of-victim factor is nearly as crucial as the statutory aggravating circumstance whether the defendant had a prior record of a conviction for a capital crime.10 Ibid. See Ga. Code Ann. § 17-10-30(b) (1982), ante, at 284-285, n. 3. The Court has noted elsewhere that Georgia could not attach “the ‘aggravating’ label to factors that are constitutionally impermissible or totally irrelevant to the sentencing process, such as for example the race, religion, or political affiliation of the defendant.” Zant v. Stephens, 462 U. S. 862, 885 (1983). What we have held to be unconstitutional if included in the *356language of the statute surely cannot be constitutional because it is a de facto characteristic of the system.
McCleskey produced evidence concerning the role of racial factors at the various steps in the decisionmaking process, focusing on the prosecutor’s decision as to which cases merit the death sentence. McCleskey established that the race of the victim is an especially significant factor at the point where the defendant has been convicted of murder and the prosecutor must choose whether to proceed to the penalty phase of the trial and create the possibility that a death sentence may be imposed or to accept the imposition of a sentence of life imprisonment. McCleskey demonstrated this effect at both the statewide level, see Supp. Exh. 56, 57, Tr. 897-910, and in Fulton County where he was tried and sentenced, see Supp. Exh. 59, 60, Tr. 978-981. The statewide statistics indicated that black-defendant/white-victim cases advanced to the penalty trial at nearly five times the rate of the black-defendant/black-victim cases (70% v. 15%), and over three times the rate of white-defendant/ black-victim cases (70% v. 19%). See Supp. Exh. 56. The multiple-regression analysis demonstrated that racial factors had a readily identifiable effect at a statistically significant level. See id., at 57; Tr. 905. The Fulton County statistics were consistent with this evidence although they involved fewer cases. See Supp. Exh. 59, 60.11
Individualized evidence relating to the disposition of the Fulton County cases that were most comparable to Mc-Cleskey’s case was consistent with the evidence of the race-of-victim effect as well. Of the 17 defendants, including *357MeCleskey, who were arrested and charged with homicide of a police officer in Fulton County during the 1973-1979 period, MeCleskey, alone, was sentenced to death. The only other defendant whose case even proceeded to the penalty phase received a sentence of life imprisonment. That defendant had been convicted of killing a black police officer. See id., at 61-63; Tr. 1050-1062.
As to the final element of the prima facie case, MeCleskey showed that the process by which the State decided to seek a death penalty in his case and to pursue that sentence throughout the prosecution was susceptible to abuse. Petitioner submitted the deposition of Lewis R. Slaton, who, as of the date of the deposition, had been the District Attorney for 18 years in the county in which MeCleskey was tried and sentenced. Deposition in No. 84-8176 of Lewis R. Slaton, Aug. 4, 1983, p. 5; see MeCleskey v. Zant, 580 F. Supp. 338, 377, n. 15 (1984); Tr. 1316. As Mr. Slaton explained, the duties and responsibilities of that office are the prosecution of felony charges within the Atlanta Judicial Circuit that comprises Fulton County. Deposition 7-8. He testified that during his years in the office, there were no guidelines informing the Assistant District Attorneys who handled the cases how they should proceed at any particular stage of the prosecution. There were no guidelines as to when they should seek an indictment for murder as opposed to lesser charges, id., at 10-11; when they should recommend acceptance of a guilty plea to murder, acceptance of a guilty plea to a lesser charge, reduction of charges, or dismissal of charges at the postindictment-preconviction stage, id., at 25-26, 31; or when they should seek the death penalty, id., at 31. Slaton testified that these decisions were left to the discretion of the individual attorneys who then informed Slaton of their decisions as they saw fit. Id., at 13, 24-25, 37-38.
Slaton’s deposition proves that, at every stage of a prosecution, the Assistant District Attorney exercised much discretion. The only guidance given was “on-the-job training.” *358Id., at 20. Addressing plea bargaining, for example, Slaton stated that “through the training that the assistant D A’s get, I think we pretty much think alike on the cases, on what we suggest.” Id., at 25. The sole effort to provide any consistency was Slaton’s periodic pulling of files at random to check on the progress of cases. Id., at 28-29. Slaton explained that as far as he knew, he was the only one aware of this checking. Id., at 28. The files contained information only as to the evidence in the case, not any indication as to why an attorney made a particular decision. The attorneys were not required to record why they sought an indictment for murder as opposed to a lesser charge, id., at 19, or why they recommended a certain plea, id., at 29-30.12 The attorneys were not required to report to Slaton the cases in which they decided not to seek the death penalty, id., at 34-36, 38, or the cases in which they did seek the death penalty, id., at 41.
When questioned directly as to how the office decided whether to seek the death penalty, Slaton listed several factors he thought relevant to that decision, including the strength of the evidence, the atrociousness of the crime, and the likelihood that a jury would impose the death sentence. Id., at 59. He explained that the attorneys did not seek the death penalty in every case in which statutory aggravating factors existed. Id., at 38-39. Slaton testified that his office still operated in the same manner as it did when he took office in 1965, except that it has not sought the death penalty in any rape cases since this Court’s decision in Coker v. Georgia, 433 U. S. 584 (1977). Deposition 60.
In addition to this showing that the challenged system was susceptible to abuse, McCleskey presented evidence of the *359history of prior discrimination in the Georgia system. Justice Brennan has reviewed much of this history in detail in his dissenting opinion, ante, at 328-334, including the history of Georgia’s racially based dual system of criminal justice. This historical background of the state action challenged “is one evidentiary source” in this equal protection case. Arlington Heights v. Metropolitan Housing Development Corp., 429 U. S. 252, 267 (1977); see also Rogers v. Lodge, 458 U. S. 613, 618, 623-625 (1982). Although I would agree that evidence of “official actions taken long ago” could not alone establish that the current system is applied in an unconstitutionally discriminatory manner, I disagree with the Court’s statement that such evidence is now irrelevant. Ante, at 298, n. 20.
The above-described evidence, considered in conjunction with the other record evidence outlined by Justice Brennan, ante, at 325-328, and discussed in opinions dissenting from the judgment of the Court of Appeals, 753 F. 2d, at 919 (Hatchett, J., dissenting in part and concurring in part); id., at 920-923 (Clark, J., dissenting in part and concurring in part), gives rise to an inference of discriminatory purpose. See Washington v. Davis, 426 U. S., at 239-242. As in the context of the rule of exclusion, see n. 6, supra, McCleskey’s showing is of sufficient magnitude that, absent evidence to the contrary, one must conclude that racial factors entered into the decisionmaking process that yielded McCleskey’s death sentence. See Castaneda v. Partida, 430 U. S., at 494, n. 13. The burden, therefore, shifts to the State to explain the racial selections. It must demonstrate that legitimate racially neutral criteria and procedures yielded this racially skewed result.
In rebuttal, the State’s expert suggested that if the Baldus thesis was correct then the aggravation level in black-victim cases where a life sentence was imposed would be higher than in white-victim cases. See 580 F. Supp., at 373. The expert analyzed aggravating and mitigating circumstances *360“one by one, demonstrating that in life sentence cases, to the extent that any aggravating circumstance is more prevalent in one group than the other, there are more aggravating features in the group of white-victim cases than in the group of black-victim cases. Conversely, there were more mitigating circumstances in which black-victim cases had a higher proportion of that circumstance than in white-victim cases.” Ibid. The District Court found that the State’s suggestion was plausible. It concluded, however, that the State did not conclusively disprove McCleskey’s case; yet it reasoned that the State’s theory “stands to contradict any prima facie case. ” Ibid. I find that reasoning wrong as a matter of law, and the conclusion clearly erroneous.
The State did not test its hypothesis to determine if white-victim and black-victim cases at the same level of aggravating circumstances were similarly treated. Tr. 1613-1614, 1664. McCleskey’s experts, however, performed this test on their data. Id., at 1297, 1729-1732, 1756-1761. They demonstrated that the racial disparities in the system were not the result of the differences in the average aggravation levels between white-victim and black-victim cases. See Supp. Exh. 72; Tr. 1291-1296; Petitioner’s Exhibit DB 92. The State’s meager and unsophisticated evidence cannot withstand the extensive scrutiny given the Baldus evidence.13 *361Here, as in Bazemore v. Friday, the State did not “demonstrate that when th[e] factors were properly organized and accounted for there was no significant disparity” between the death sentences imposed on defendants convicted of killing white victims and those imposed on defendants convicted of killing black victims. 478 U. S., at 403-404, n. 14. In Castaneda, we rejected a similar effort by the State to rely on an unsupported countervailing theory to rebut the evidence. 430 U. S., at 500. In sum, McCleskey has demonstrated a clear pattern of differential treatment according to race that is “unexplainable on grounds other than race.” Arlington Heights v. Metropolitan Housing Development Corp., 429 U. S., at 266.
Ill
The Court’s explanations for its failure to apply this well-established equal protection analysis to this case are not persuasive. It first reasons that “each particular decision to impose the death penalty is made by a petit jury” and that 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-selection or Title VII *362case.” Ante, at 294-295. According to the Court, the statistical evidence is less relevant because, in the two latter situations, there are fewer variables relevant to the decision and the “statistics relate to fewer entities.” Ante, at 295.
I disagree with the Court’s assertion that there are fewer variables relevant to the decisions of jury commissioners or prosecutors in their selection of jurors, or to the decisions of employers in their selection, promotion, or discharge of employees. Such decisions involve a multitude of factors, some rational, some irrational. Second, I disagree with the comment that the venire-selection and employment decisions are “made by fewer entities.” Certainly in the employment context, personnel decisions are often the product of several levels of decisionmaking within the business or government structure. The Court’s statement that the decision to impose death is made by the petit jury also disregards the fact that the prosecutor screens the cases throughout the pretrial proceedings and decides to seek the death penalty and to pursue a capital case to the penalty phase where a death sentence can be imposed. McCleskey’s claim in this regard lends itself to analysis under the framework we apply in assessing challenges to other prosecutorial actions. See Batson v. Kentucky, 476 U. S. 79 (1986); see also Wayte v. United States, 470 U. S. 598, 608, n. 10 (1985) (applying Castaneda framework in challenge to prosecutor’s allegedly selective enforcement of criminal sanction). It is appropriate to judge claims of racially discriminatory prosecutorial selection of cases according to ordinary equal protection standards. 470 U. S., at 608.
The Court’s other reason for treating this case differently from venire-selection and employment cases is that in these latter contexts, “the decisionmaker has an opportunity to explain the statistical disparity,” but in the instant case the State had no practical opportunity to rebut the Baldus study. Ante, at 296. According to the Court, this is because jurors cannot be called to testify about their verdict and because *363policy considerations render it improper to require “prosecutors to defend their decisions to seek death penalties, ‘often years after they were made.’” Ibid., quoting Imbler v. Pachtman, 424 U. S. 409, 425 (1976).
I agree with the Court’s observation as to the difficulty of examining the jury’s decisionmaking process. There perhaps is an inherent tension between the discretion accorded capital sentencing juries and the guidance for use of that discretion that is constitutionally required. In his dissenting opinion, Justice Brennan demonstrates that the Eighth Amendment analysis is well suited to address that aspect of the case. Ante, at 323. The Court’s refusal to require that the. prosecutor provide an explanation for his actions, however, is completely inconsistent with this Court’s longstanding precedents. The Court misreads Imbler v. Pachtman. In that case, the Court held that a prosecutor who acted within the scope of his duties was entitled to absolute immunity in an action under 42 U. S. C. § 1983 for damages. We recognized that immunity from damages actions was necessary to prevent harassing litigation and to avoid the threat of civil litigation undermining the prosecutor’s independence of judgment. We clearly specified, however, that the policy considerations that compelled civil immunity did not mean that prosecutors could not be called to answer for their actions. We noted the availability of both criminal sanctions and professional ethical discipline. 424 U. S., at 429. Prosecutors undoubtedly need adequate discretion to allocate the resources of their offices and to fulfill their responsibilities to the public in deciding how best to enforce the law, but this does not place them beyond the constraints imposed on state action under the Fourteenth Amendment. Cf. Ex parte Virginia, 100 U. S. 339 (1880) (upholding validity of conviction of state judge for discriminating on the basis of race in his selection of jurors).
The Court attempts to distinguish the present case from Batson v. Kentucky, in which we recently reaffirmed the fact *364that prosecutors’ actions are not unreviewable. See ante, at 296, n. 17. I agree with the Court’s observation that this case is “quite different” from the Batson case. Ibid. The irony is that McCleskey presented proof in this case that would have satisfied the more burdensome standard of Swain v. Alabama, 380 U. S. 202 (1965), a standard that was described in Batson as having placed on defendants a “crippling burden of proof.” 476 U. S., at 92. As discussed above, McCleskey presented evidence of numerous decisions imper-missibly affected by racial factors over a significant number of cases. The exhaustive evidence presented in this case certainly demands an inquiry into the prosecutor’s actions.
The Court’s assertion that, because of the necessity of discretion in the criminal justice system, it “would demand exceptionally clear proof,” ante, at 297, before inferring abuse of that discretion thus misses the point of the constitutional challenge in this case. Its conclusory statement that “the capacity of prosecutorial discretion to provide individualized justice is ‘firmly entrenched in American law,’” ante, at 311-312, quoting 2 W. LaFave & J. Israel, Criminal Procedure § 13.2(a), p. 160 (1984), is likewise not helpful. The issue in this case is the extent to which the constitutional guarantee of equal protection limits the discretion in the Georgia capital sentencing system. As the Court concedes, discretionary authority can be discriminatory authority. Ante, at 312. Prosecutorial decisions may not be “‘deliberately based upon an unjustifiable standard such as race, religion, or other arbitrary classification.’” Bordenkircher v. Hayes, 434 U. S. 357, 364 (1978), quoting Oyler v. Boles, 368 U. S. 448, 456 (1962). Judicial scrutiny is particularly appropriate in McCleskey’s case because “[m]ore subtle, less consciously held racial attitudes could also influence” the decisions in the Georgia capital sentencing system. Turner v. Murray, 476 U. S. 28, 35 (1986); see n. 13, supra. The Court’s rejection of McCleskey’s equal protection claims is *365a far cry from the “sensitive inquiry” mandated by the Constitution.
IV
A
One of the final concerns discussed by the Court may be the most disturbing aspect of its opinion. Granting relief to McCleskey in this case, it is said, could lead to further constitutional challenges. Ante, at 314-319. That, of course, is no reason to deny McCleskey his rights under the Equal Protection Clause. If a grant of relief to him were to lead to a closer examination of the effects of racial considerations throughout the criminal justice system, the system, and hence society, might benefit. Where no such factors come into play, the integrity of the system is enhanced. Where such considerations are shown to be significant, efforts can be made to eradicate their impermissible influence and to ensure an evenhanded application of criminal sanctions.
B
Like Justice Stevens, I do not believe acceptance of McCleskey’s claim would eliminate capital punishment in Georgia. Post, at 367. Justice Stevens points out that the evidence presented in this case indicates that in extremely aggravated murders the risk of discriminatory enforcement of the death penalty is minimized. Ibid. I agree that narrowing the class of death-eligible defendants is not too high a price to pay for a death penalty system that does not discriminate on the basis of race. Moreover, the establishment of guidelines for Assistant District Attorneys as to the appropriate basis for exercising their discretion at the various steps in the prosecution of a case would provide at least a measure of consistency. The Court’s emphasis on the procedural safeguards in the system ignores the fact that there are none whatsoever during the crucial process leading up to trial. As Justice White stated for the plurality in Turner v. Murray, I find “the risk that racial prejudice may *366have infected petitioner’s capital sentencing unacceptable in light of the ease with which that risk could have been minimized.” 476 U. S., at 36. I dissent.
I agree with Justice Stevens’ position that the proper course is to remand this case to the Court of Appeals for determination of the validity of the statistical evidence presented. Post, at 367. Like Justice Stevens, however, I am persuaded that the Baldus study is valid and would remand merely in the interest of orderly procedure.
See, e. g., H. R. Joint Comm. Rep. No. 30, 39th Cong., 1st Sess., pt. II, p. 25 (1866) (testimony of George Tucker, Virginia attorney) (“They have not any idea of prosecuting white men for offenses against colored people; they do not appreciate the idea”); id., at 209 (testimony of Dexter H. Clapp) (“Of the thousand cases of murder, robbery, and maltreatment of freedmen that have come before me, ... I have never yet known a single ease in which the local authorities or police or citizens made any attempt or exhibited any inclination to redress any of these wrongs or to protect such persons”); id., at 213 (testimony of J. A. Campbell) (although identities of men suspected of killing two blacks known, no arrest or trial had occurred); id., pt. Ill, p. 141 (testimony of Brev. Maj. Gen. Wager Swayne) (“I have not known, after six months’ residence at the capital of the State, a single instance of a white man being convicted and hung or sent to the penitentiary for crime against a negro, while many cases of crime warranting such punishment have been reported to me”); id., pt. IV, p. 75 (testimony of Maj. Gen. George A. Custer) (“[I]t is of weekly, if not of daily, occurrence that freedmen are murdered. . . . [Sjometimes it is not known who the perpetrators are; but when that is known no action is taken against them. I believe a white man has never been hung for murder in Texas, although it is the law”).
In Brown v. Board of Education, 347 U. S. 483 (1954), this Court held that, despite the fact that the legislative history of the Fourteenth Amendment indicated that Congress did not view racial discrimination in public education as a specific target, the Amendment nevertheless prohibited such discrimination. The Court today holds that even though the Fourteenth Amendment was aimed specifically at eradicating discrimination in the enforcement of criminal sanctions, allegations of such discrimination supported by substantial evidence are not constitutionally cognizable. But see Batson v. Kentucky, 476 U. S. 79, 85 (1986) (allegations of racially discriminatory exercise of peremptory challenges by prosecutor subject to review under Fourteenth Amendment because “[ejxclusion of black citizens from service as jurors constitutes a primary example of the evil the Fourteenth Amendment was designed to cure”).
The Court refers to the prosecutor’s role in the capital sentencing process without analyzing the import of the statistical evidence concerning the steps of the process at which the prosecutor determines the future of the case. The Court recognizes that the prosecutor determines whether a case even will proceed to the penalty phase. If the prosecutor does not pursue the death penalty, a mandatory sentence of life imprisonment is imposed. See ante, at 284, n. 2. It lists many of the factors that prosecutors take into account in making their decisions, ante, at 307-308, n. 28, and recognizes that in each case the prosecutor can decline to charge, or to offer a plea bargain, or to seek a death sentence, ante, at 312. It also notes that the Baldus study “found that prosecutors sought the death penalty in 70% of the cases involving black defendants and white victims; 32% *351of the cases involving white defendants and white victims; 15% of the cases involving black defendants and black victims; and 19% of the eases involving white defendants and black victims,” ante, at 287.
The Court relies heavily on its assertion that prosecutorial discretion should not be reviewed, ante, at 296-297, 311-312, but elsewhere concedes that such discretion may not be exercised in a racially discriminatory manner, ante, at 309, n. 30. It nowhere explains why this limitation on pros-ecutorial discretion does not require the same analysis that we apply in other cases involving equal protection challenges to the exercise of pros-ecutorial discretion. See, e. g., Batson v. Kentucky, 476 U. S. 79 (1986).
The use of the prima facie case method to structure proof in eases charging racial discrimination is appropriate because it “progressively . . . *352sharpen[s] the inquiry into the elusive factual question of intentional discrimination.” Texas Dept. of Community Affairs v. Burdine, 450 U. S. 248, 255, n. 8 (1981); see McCleskey v. Kemp, 753 F. 2d 877, 912 (CA11 1985) (Johnson, J., dissenting in part and concurring in part) (where the “prosecutor has considerable discretion and the jury has bounded but irreducible discretion,” the discretion could easily mask conscious or unconscious racial discrimination and indirect methods of proof are therefore required as outlined in Washington v. Davis, 426 U. S. 229, 241-242 (1976), and Arlington Heights v. Metropolitan Housing Development Corp., 429 U. S. 252, 266, n. 13 (1977)).
The Court recently explained: “In deciding if the defendant has carried his burden of persuasion, a court must undertake ‘a sensitive inquiry into such circumstantial and direct evidence of intent as may be available.’ Arlington Heights v. Metropolitan Housing Development Corp., 429 U. S., at 266. Circumstantial evidence of invidious intent may include proof of disproportionate impact. Washington v. Davis, 426 U. S., at 242. We have observed that under some circumstances proof of .discriminatory impact ‘may for all practical purposes demonstrate unconstitutionality because in various circumstances the discrimination is very difficult to explain on nonraeial grounds.’ Ibid.’’ Batson v. Kentucky, 476 U. S., at 93.
In Castaneda, we explained that in jury-selection cases where the criminal defendant is attempting to prove that there was discriminatory *353exclusion of potential jurors we apply the “rule of exclusion” method of proof. 430 U. S., at 494. The underlying rationale is that “[i]f a disparity is sufficiently large, then it is unlikely that it is due solely to chance or accident, and, in the absence of evidence to the contrary, one must conclude that racial or other class-related factors entered into the selection process.” Id., at 494, n. 13.
Although the Court states that it assumes the validity of the Baldus study for purposes of its analysis, because of its detailed discussion of the District Court’s reasons for rejecting its validity I am compelled to record my disagreement with the District Court’s reasoning. As a member of the United States Court of Appeals, I was confronted in 1968 with a challenge to the constitutionality of a State’s capital sentencing system based on allegations of racial discrimination supported by statistical evidence. Writing for a panel of the court, I rejected that challenge for reasons similar to those espoused by the Court today. Maxwell v. Bishop, 398 F. 2d 138 (CA8), vacated and remanded, sua sponte, on grounds not raised below, 398 U. S. 262 (1970) (per curiam).
The Court of Appeals found the evidence presented by Maxwell incomplete, not directly relevant to his individual claim, and statistically insufficient. McCleskey’s evidence, however, is of such a different level of sophistication and detail that it simply cannot be rejected on those grounds. Unlike the evidence presented by Maxwell, which did not contain data from the jurisdiction in which he was tried and sentenced, McCleskey’s evidence includes data from the relevant jurisdiction. Whereas the analyses presented by Maxwell did not take into account a significant number of variables and were based on a universe of 55 cases, the analyses presented by McCleskey’s evidence take into account more than 400 variables and áre based on data concerning all offenders arrested for homicide in Georgia from 1973 through 1978, a total of 2,484 cases. Moreover, the sophistication of McCleskey’s evidence permits consideration of the existence of racial discrimination at various decision points in the process, not merely at the jury decision. It is this experience, in part, that convinces me of the significance of the Baldus study.
See Brief for Dr. Franklin M. Fisher et al. as Amici Curiae 19.
A defendant’s chances of receiving a death sentence increase by a factor of 4.3 if the victim is white, but only by 2.3 if the defendant was the prime mover behind the homicide.
A prior record of a conviction for murder, armed robbery, rape, or kidnaping with bodily injury increases the chances of a defendant’s receiving a death sentence by a factor of 4.9.
The universe of eases from Fulton County analyzed by Baldus included 629 killings, 581 of which yielded murder indictments. Supp. Exh. 59, 60; Tr. 978-981. The evidence indicated that at each step in the process from indictment to sentence, there is a differential treatment in the disposition of white-victim and black-victim cases, with the white-victim cases having a higher likelihood of being retained in the system and risking a death sentence. Supp. Exh. 60; Tr. 978-981.
In his deposition, Russell Parker, the Assistant District Attorney who prosecuted McCleskey’s case, contradicted the statement cited by the Court, ante, at 312, n. 34, concerning plea negotiations during Mc-Cleskey’s trial. Parker testified that he never discussed a plea with McCleskey. Deposition in No. 84-8176 of Russell Parker, Feb. 16, 1981, p. 15.
As a result of MeCleskey’s discovery efforts, the record also contains relevant testimonial evidence by two state officials. The Fulton County District Attorney testified that he did not recall any instance in which race was a factor in a death penalty case in his office. Deposition in No. 84-8176 of Lewis R. Slaton, Aug. 4, 1983, p. 78. He later recalled one ease that was in the office when he first began, in which the office set aside the death penalty because of the possibility that race had been involved. Id., at 79-80. The Assistant District Attorney who prosecuted McCleskey’s case testified that race did not influence his decision to seek the death penalty in the present case. Deposition of Russell Parker, Feb. 16, 1981, p. 17.
These general assertions by state officials that they did not discriminate or that they properly performed their official duties, however, cannot meet *361the State’s burden of rebuttal of the prima facie case. See Alexander v. Louisiana, 406 U. S. 625, 631-632 (1972); Whitus v. Georgia, 385 U. S. 545, 551-552 (1967). Moreover, there are many ways in which racial factors can enter indirectly into prosecutorial decisions. For example, the authors of a study similar to that of Baldus explained: “Since death penalty prosecutions require large allocations of scarce prosecutorial resources, prosecutors must choose a small number of eases to receive this expensive treatment. In making these choices they may favor homicides that are visible and disturbing to the majority of the community, and these will tend to be white-victim homicides.” Gross & Mauro, Patterns of Death: An Analysis of Racial Disparities in Capital Sentencing and Homicide Victimization, 37 Stan. L. Rev. 27, 106-107 (1984); see generally Johnson, Race and the Decision to Detain a Suspect, 93 Yale L. J. 214 (1983); Lawrence, The Id, the Ego, and Equal Protection: Reckoning with Unconscious Racism, 39 Stan. L. Rev. 317 (1987).