with whom Justice Marshall joins, dissenting.
Today’s decision is a conspicuous exercise in judicial activism — particularly so since it takes the form of disregard of precedent scarcely a month old. In its eagerness to expatiate upon the “significant costs” of the Great Writ, ante, at 126-128, and to apply “the principles articulated in Wainwright v. Sykes, [433 U. S. 72 (1977)],” ante, at 123, to the cases before us, the Court demonstrably misreads and reshapes the habeas claim of at least one of the state prisoners involved in this action. Respondent Isaac presented exactly one claim in his habeas petition. That claim did not even *138exist until after Isaac was denied relief on his last direct appeal. As a result, Isaac could not have “preserved” his claim in the state courts: He simply committed no “procedural default,” and the Court is thus clearly wrong to apply Sykes to his claim in order to relegate it to the dustbin. Moreover, the Court does so by ignoring the holding only last month in Rose v. Lundy, 455 U. S. 509 (1982): namely, that a habeas petition that contains any unexhausted claims must be dismissed by the habeas court. The Court then compounds its error when it attempts to articulate the “principles” of Sykes: In purporting to give content to the “cause” standard announced in that case, the Court defines “cause” in a way supported neither by Sykes nor by common sense. I dissent from both of these errors, which are discussed in turn below.
HH
Respondent Isaac was indicted m May 1975; he was convicted after a jury trial and sentenced during the following September.1 While his conviction was on appeal in the Ohio Court of Appeals, the Ohio Supreme Court decided State v. Robinson, 47 Ohio St. 2d 103, 351 N. E. 2d 88 (July 1976), which construed Ohio Rev. Code Ann. § 2901.05(A) (effective Jan. 1, 1974) to require the prosecution to bear the burden of persuasion, beyond a reasonable doubt, with respect to an affirmative defense of self-defense raised by the defendant. The Ohio Court of Appeals affirmed Isaac’s conviction in February 1977.2 The Ohio Supreme Court dismissed Isaac’s appeal in July 1977.3 On the same day, the Ohio Supreme Court decided State v. Humphries, 51 Ohio St. 2d 95, 364 N. E. 2d 1354. That case declared Robinson retroactive to the effective date of § 2901.05(A), but only partially: It held that in order to gain the retroactive benefits of the Robinson *139decision, a defendant tried before a jury must have preserved his claim by objection at trial to the allocation of the affirmative-defense burden of proof, while a bench-trial defendant could have made the same objection as late as in the Court of Appeals, and the objection would still have been preserved. 51 Ohio St. 2d, at 102-103, 364 N. E. 2d, at 1359.
Isaac filed his habeas petition in the United States District Court for the Southern District of Ohio in March 1978.4 The asserted ground for relief was “denial of due process of law,” in that
“[t]he trial court charged petitioner had the burden of proving self-defense. After conviction and during the first appeal the Ohio Supreme Court declared the instructions to be prejudicial error under Robinson. This case was immediately raised to the Appellate Court. They held any error was waived. The Ohio Supreme Court then held Robinson retroactive. Petitioner had raised retroactivity in its leave to appeal and was denied leave to appeal the same day Humphries was decided declaring retroactivity. The Ohio Supreme Court refuses to give relief despite its own pronouncement. The holding of the court is contrary to the Supreme Court of the United States in regard to proving self-defense.”5
Isaac’s memorandum in support of his habeas petition made it plain that his claim was that Humphries’ selective retroactive application of the Robinson rule denied him due process of law.6 It is obvious, of course, that it was simply impossible *140to make this claim before Humphries was decided, in July-1977, on the same day that Isaac’s direct appeals in the state court system were finally rejected.
Ohio Rev. Code Ann. § 2958.21(A) (1975) provides for post-conviction relief under certain circumstances:
“Any person convicted of a criminal offense . . . claiming that there was such a denial or infringement of his rights as to render the judgment void or voidable under the Ohio Constitution or the Constitution of the United States, may file a verified petition at any time in the court which imposed sentence, stating the grounds for relief relied upon, and asking the court to vacate or set aside the judgment or sentence or to grant other appropriate relief.”
By applying the doctrine of res judicata to postconviction petitions, the Ohio Supreme Court has allowed relief under this procedure only under limited circumstances: Constitutional issues can be raised under § 2953.21(A) only when they could not have been raised at trial or on appeal. State v. Perry, 10 Ohio St. 2d 175, 180-181, 226 N. E. 2d 104, 108 (1967); see Keener v. Ridenour, 594 F. 2d 581, 589-591 (CA6 1979) (construing scope of Ohio postconviction remedy); Riley v. Havener, 391 F. Supp. 1177, 1179-1180 (ND Ohio 1974) (same). But Isaac’s claim is manifestly of the sort that could not have been raised at trial or on appeal, for the claim only came into existence on the day that Isaac’s last appeal was rejected. Consequently, state postconviction remedies are available to Isaac and have not been exhausted.
I draw three conclusions from the foregoing account, all of which to my mind follow ineluctably from the undisputed facts of this case. . First, Isaac’s habeas petition should have been dismissed for his failure to exhaust available state remedies. See Picard v. Connor, 404 U. S. 270 (1971), where we emphasized that
“the federal claim must be fairly presented to the state courts. . . . Only if the state courts have had the first *141opportunity to hear the claim sought to be vindicated in a federal habeas proceeding does it make sense to speak of the exhaustion of state remedies.” Id., at 275-276.
In the present case, petitioner Engle responded to Isaac’s petition by raising the issue of Isaac’s failure to exhaust.7 Therefore the Court of Appeals clearly erred, under Picard and our whole line of exhaustion precedents, in granting ha-beas relief to Isaac instead of requiring exhaustion. The proper disposition of Isaac’s case is thus to reverse and remand with instructions to dismiss on exhaustion grounds. The Court’s failure to order such a disposition is incomprehensible: Barely a month ago this Court emphatically reaffirmed the exhaustion doctrine, and indeed extended it, announcing a requirement of “total exhaustion” for habeas petitions. Rose v. Lundy, 455 U. S. 509 (March 3, 1982).8 But today the Court finds the nostrum of “cause and prejudice” more attractive, and so Rose v. Lundy is not applied. Sic transit gloria Lundy! In scarcely a month, the bloom is off the Rose.9
My second conclusion is that Isaac simply committed no “procedural default” in failing to raise at trial or on direct appeal the claim that appears in his habeas petition. That claim did not exist at any time during Isaac’s trial or direct appeal. Thus the essential factual predicate for an application of Wainwright v. Sykes, 433 U. S. 72 (1977), is com*142pletely absent in Isaac’s case. Sykes involved a habeas petitioner who had failed to object in a timely manner to the admission of his confession at trial. Id., at 86-87. Given that factual predicate, Sykes addressed the question of whether federal habeas review should be barred absent a showing of “cause” for the procedural default of failing to object, and a further showing of “prejudice” resulting from the admission of the confession. Id., at 87, 90-91. But in the case before us, respondent Isaac could not have made any objection, timely or otherwise, at trial or on appeal. Thus the application of Sykes is completely and manifestly erroneous in this case.10
My last conclusion is that the Court is so intent upon applying Sykes to Isaac’s case that it plays Procrustes with his claim. In order to bring Isaac’s claim within the ambit of Sykes, the Court first characterizes his petition as “complex,” ante, at 117, and “confused,” ante, at 124, n. 25.11 Then, *143without ever quoting the claim as it actually appeared in Isaac’s petition, the Court delineates a “colorable constitutional claim” nowhere to be found in the petition. As the Court recasts it, Isaac’s claim is as follows:
“[T]he crim[e] charged against [Isaac] require[s] a showing of purposeful or knowing behavior. These terms, according to [Isaac], imply a degree of culpability that is absent when a person acts in self-defense. . . . Self-defense, [Isaac] urge[s], negates [essential] elements of criminal behavior. Therefore, once the defendant raises the possibility of self-defense, [Isaac] contend[s] that the State must disprove that defense as part of its task of establishing guilty mens rea, voluntariness, and unlawfulness. The Due Process Clause, according to [Isaac’s] interpretation of Winship, Mullaney, and Patterson, forbids the States to disavow any portion of this burden.” Ante, at 121-122.
This new-modeled claim bears no resemblance to the claim actually made by Isaac in his habeas petition. See supra, at 139.12 But by virtue of this exercise in juristic revisionism, the Court puts itself in position to find that “Isaac’s” claim was “forfeited before the state courts,” ante, at 125 — no difficult task, since the claim is wholly imagined by the Court itself — thus enabling the Court to reach its clearly sought goal of deciding “whether the principles articulated in Wainwright v. Sykes, 433 U. S. 72 (1977), bar consideration of the claim in a federal habeas proceeding.” Ante, at 123. Unsurprisingly, the Court’s bottom line is that Isaac’s Active claim is indeed barred by Sykes. In short, the Court reshapes respondent Isaac’s actual claim into a form that enables it to foreclose all federal review, when as plainly pleaded the claim was unexhausted, thus calling for the dismissal of Isaac’s pe*144tition for habeas relief. The Court’s analysis is completely result-oriented, and represents a noteworthy exercise in the very judicial activism that the Court so deprecates in other contexts.
II
For the reasons stated above, I conclude that in its unseemly rush to reach the merits of Isaac’s case, the Court has ignored settled law respecting the exhaustion of state remedies. But lest it be thought that my disagreement with today’s decision is confined to that point alone, I turn to the Court’s treatment of the merits of the cases before us. I continue to believe that the “deliberate bypass” standard announced in Fay v. Noia, 372 U. S. 391 (1963), is the only sensible rule to apply in habeas cases such as respondents’. I adhere to my dissent in Wainwright v. Sykes, supra, in which I termed the “cause-and-prejudice” standard adopted in that case “a mere house of cards whose foundation has escaped any systematic inspection.” 433 U. S., at 99-100, n. 1. The Court has now begun to furnish its house of cards — and the furniture is as jerry-built as the house itself.
A
Sykes did not give the terms “cause” and “prejudice” any “precise content,” but promised that “later cases” would provide such content. Id., at 91. Today the nature of that content becomes distressingly apparent. The Court still refuses to say what “cause” is: And I predict that on the Court’s present view it will prove easier for a camel to go through the eye of a needle than for a state prisoner to show “cause.” But on the other hand, the Court is more than eager to say what “cause” is not: And in doing so, the Court is supported neither by common sense nor by the very reasons offered in Sykes for adoption of the “cause-and-prejudice” standard in the first place.
According to the Court, “cause” is not demonstrated when the Court “cannot say that [habeas petitioners] lacked the *145tools to construct their constitutional claim,” ante, at 133, however primitive those tools were and thus however inchoate the claim was when petitioners were in the state courts. The Court concludes, after several pages of tortuous reasoning, ante, at 130-133, and nn. 36-42, that respondents in the present cases did indeed have “the tools” to make their constitutional claims. This conclusion is reached by the sheerest inference: It is based on citations to other cases in other jurisdictions, where other defendants raised other claims assertedly similar to those that respondents “could” have raised. Ante, at 131-133, and n. 40. To hold the present respondents to such a high standard of foresight is tantamount to a complete rejection of the notion that there is a point before which a claim is so inchoate that there is adequate “cause” for the failure to raise it. In thus rejecting in-choateness as “cause,” the Court overlooks the fact that none of the rationales used in Sykes to justify adoption of the cause-and-prejudice standard can justify today’s definition of “cause.”
Sykes adopted the cause-and-prejudice standard in order to accord “greater respect” to state contemporaneous-objection rules than was assertedly given by Fay v. Noia, supra. 433 U. S., at 88. The Court then offered a number of reasons why contemporaneous-objection rules should be given such greater respect:
(1) “A contemporaneous objection enables the record to be made with respect to the constitutional claim when the recollections of witnesses are freshest, not years later in a federal habeas proceeding.” Ibid.
(2) A contemporaneous objection “enables the judge who observed the demeanor of those witnesses to make the factual determinations necessary for properly deciding the federal constitutional question.” Ibid.
(3) “A contemporaneous-objection rule may lead to the exclusion of evidence objected to, thereby making a major contribution to finality in criminal litigation.” Ibid.
*146(4) The Fay v. Noia rule “may encourage ‘sandbagging’ on the part of defense lawyers, who may take their chances on a verdict of not guilty in a state trial court with the intent to raise their constitutional claims in a federal habeas court if their initial gamble does not pay off.” 433 U. S., at 89.
(5) A contemporaneous-objection rule “encourages the result that [criminal trials] be as free of error as possible.” Id., at 90.
None of these rationales has any force in the present case. The first three reasons are valid, if at all, only in the particular context of objections to the admission of evidence, such as were at issue in Sykes. As for the “sandbagging” rationale, dutifully repeated by today’s Court, ante, at 129, n. 34, that was fully answered in my Sykes dissent:13 That argument still “offends common sense,” and does not become less offensive by sententious repetition. And the final reason — relied on again today, ante, at 127—is plainly irrelevant to a case involving inchoate constitutional claims. Such claims are ex hypothesis so embryonic that only the extraordinarily foresighted criminal defendant will raise them. It is completely implausible to expect that the raising of such claims will predictably — or even occasionally — make trials more “free of error.”
B
The Court justifies its result today with several additional reasons — or, rather, sentiments in reasons’ clothing. We are told, ante, at 126-127, that “the Great Writ entails sig*147nificant costs. Collateral review of a conviction extends the ordeal of trial for both society and the accused.” But we are not told why the accused would consider it an “ordeal” to go to federal court in order to attempt to vindicate his constitutional rights. Nor are we told why society should be eager to ensure the finality of a conviction arguably tainted by un-reviewed constitutional error directly affecting the truth-finding function of the trial. I simply fail to understand how allowance of a habeas hearing “entails significant costs” to anyone under the circumstances of the cases before us.
In a similar vein, we are told, ante, at 127, that “[w]e must also acknowledge that writs of habeas corpus frequently cost society the right to punish admitted offenders.” I for one will acknowledge nothing of the sort. Respondents were all convicted after trials in which they allege that the burden of proof respecting their affirmative defenses was imposed upon them in an unconstitutional manner. Thus they are not “admitted” offenders at all: If they had been tried with the assertedly proper allocation of the burden of proof, then they might very well have been acquitted. Further, it is sheer demagoguery to blame the “offender” for the logistical and temporal difficulties arising from retrial: If the writ of habeas *148corpus has been granted, then it is at least as reasonable to blame the State for having prosecuted the first trial “in violation of the Constitution or laws ... of the United States,” 28 U. S. C. § 2254(a).
Finally, we are told that
“the Great Writ imposes special costs on our federal system”; that “[fjederal intrusions into state criminal trials frustrate both the States’ sovereign power to punish offenders and their good-faith attempts to honor constitutional rights,” ante, at 128; and that “[s]tate courts are understandably frustrated when they faithfully apply existing constitutional law only to have a federal court discover, during a § 2254 proceeding, new constitutional commands.” Ante, at 128, n. 33.
Once again, the Court drags a red herring across its path. I hope that the Court forgets only momentarily that “the States’ sovereign power” is limited by the Constitution of the United States: that the “intrusion” complained of is that of the supreme law of the land. But it must be reason for deep concern when this Court forgets, as it certainly does today, that “it is a constitution we are expounding, ... a constitution intended to endure for ages to come, and, consequently, to be adapted to the various crises of human affairs.”14 It is inimical to the principle of federal constitutional supremacy to defer to state courts’ “frustration” at the requirements of federal constitutional law as it is interpreted in an evolving society. Sykes promised that its cause-and-prejudice standard would “not prevent a federal habeas court from adjudicating for the first time the federal constitutional claim of a defendant who in the absence of such an adjudication will be the victim of a miscarriage of justice.” 433 U. S., at 91. Today’s decision, with its unvarnished hostility to the assertion of federal constitutional claims, starkly reveals the emptiness of that promise.
*149C
Finally, there is the issue of the Court’s extension of the Sykes standard “to cases in which the constitutional error . . . affectfs] the truthfinding function of the trial.” Ante, at 129. The Court concedes, ibid., that Sykes itself involved the violation of the habeas petitioner’s Miranda rights, and that although “this defect was serious, it did not affect the determination of guilt at trial.” But despite the fact that the present cases admittedly do involve a defect affecting the determination of guilt, the Court refuses to limit Sykes and thus bars federal review: “We do not believe . . . that the principles of Sykes lend themselves to this limitation.” Ante, at 129. In so holding, the Court ignores the manifest differences between claims that affect the truthfinding function of the trial and claims that do not.
The Court proclaimed in Stone v. Powell, 428 U. S. 465, 490 (1976), that “the ultimate question of guilt or innocence . . . should be the central concern in a criminal proceeding.” A defendant’s Fourth Amendment rights, see Stone, or his Miranda rights, see Sykes, may arguably be characterized as “crucially different from many other constitutional rights,” Kaufman v. United States, 394 U. S. 217, 237 (1969) (Black, J., dissenting), in that evidence procured in violation of those rights has not ordinarily been rendered untrustworthy by the means of its procurement. But a defendant’s right to a trial at which the burden of proof has been constitutionally allocated can never be violated without rendering the entire trial result untrustworthy. “In all kinds of litigation it is plain that where the burden of proof lies may be decisive of the outcome,” Speiser v. Randall, 357 U. S. 513, 525 (1958), and petitioners in the present cases concede as much, Brief for Petitioners 22. As Justice Harlan noted in In re Winship, 397 U. S. 358 (1970):
“If, for example, the standard of proof for a criminal trial were a preponderance of the evidence rather than proof beyond a reasonable doubt, there would be a smaller risk *150of factual errors that result in freeing guilty persons, but. a far greater risk of factual errors that result in convicting the innocent.” Id., at 371 (concurring opinion).
Where, as here, the burden was placed on respondents, rather than on the prosecution, to prove their affirmative defenses by a preponderance of the evidence, the risk of convicting the innocent is even greater than in Justice Harlan’s example. And if this allocation of the burden of proof was erroneous, then that error constitutes a denial of due process of intolerable proportions. We have recognized the truth of this proposition in numerous precedents. In Ivan V. v. City of New York, 407 U. S. 203 (1972), we held our earlier decision in Winship to be fully retroactive, stating:
“‘Where the major purpose of a new constitutional doctrine is to overcome an aspect of a criminal trial that substantially impairs its truth-finding function and so raises serious questions about the accuracy of guilty verdicts in past trials, the new rule has been given complete retroactive effect. Neither good-faith reliance by state or federal authorities on prior constitutional law or accepted practice, nor severe impact on the administration of justice has sufficed to require prospective application in these circumstances.’ Williams v. United States, 401 U. S. 646, 653 (1971). See Adams v. Illinois, 405 U. S. 278, 280 (1972); Roberts v. Russell, 392 U. S. 293, 295 (1968).” 407 U. S., at 204 (emphasis added).15
In sum, this Court has heretofore adhered to the principle that “[i]n the administration of criminal justice, our society imposes almost the entire risk of error upon itself,” because “the interests of the defendant are of such magnitude.” Addington v. Texas, 441 U. S. 418, 423-424 (1979). In the *151context of the cases before us today, this principle means that a habeas claim that a mistake was made in imposing that risk of error cannot be cavalierly dismissed as just another “type of claim raised by the prisoner,” ante, at 129. In my view, the Sykes standard is misguided and insupportable in any context. But if it is to be suffered to exist at all, it should be limited to the arguable peripheries of the trial process: It should not be allowed to insulate from all judicial review all violations of the most fundamental rights of the accused.
I dissent.
App. 2; App. to Brief in No. 78-3488 (CA6), pp. 2, 3-4.
App. 6.
Id., at 13.
App. to Brief in No. 78-3488 (CA6), p. 18.
Id., at 21 (emphasis added).
Id., at 25: “[T]he Ohio Supreme Court denied [Isaac] leave to appeal on the same day it decided State v. Humphries, . . . which declared its ruling in Robinson to be retroactive to January 1, 1974. . . . [Isaac] submits to make Robinson retroactive, and then to refuse to give him the benefits of retroactivity violates the due process guarantees of the Fourteenth Amendment. . .
Id., at 35-36.
“A rigorously enforced total exhaustion rule will encourage state prisoners to seek full relief first from the state courts, thus giving those courts the first opportunity to review all claims of constitutional error. As the number of prisoners who exhaust all of their federal claims increases, state courts may become increasingly familiar with and hospitable toward federal constitutional issues.” 455 U. S., at 518-519.
The Court notes, ante, at 123-124, n. 25, that Isaac added citations to Mullaney v. Wilbur, 421 U. S. 684 (1975), and Patterson v. New York, 432 U. S. 197 (1977), in his memorandum in support of his habeas petition. The Court apparently holds that these citations somehow save Isaac’s petition from dismissal. But that holding is flatly contrary to the explicit holding of Rose, that “the exhaustion rule in 28 U. S. C. §§ 2254(b), (c) requires *142a federal district court to dismiss a petition for a writ of habeas corpus containing any claims that have not been exhausted in the state courts.” 455 U. S., at 510 (emphasis added).
Recognizing this flat contradiction, the Court suggests that the claim “touted” by me “formed no part of Isaac’s original habeas petition.” Ante, at 124, n. 25. This suggestion is clearly belied by the plain language of Isaac’s habeas petition, which the Court never quotes, but which is quoted in full supra, at 139. That language speaks for itself, far more clearly and eloquently than the Court’s unsuccessful attempt to reconstruct it.
The panel opinion of the United States Court of Appeals for the Sixth Circuit in Isaac’s case reached this same conclusion. The panel correctly read Isaac’s petition as presenting the question of “whether the decision of the Supreme Court of Ohio to withhold from petitioner the benefits of Section 2901.05(A), as established in State v. Robinson, for failure to comply with Ohio’s contemporaneous objection rule was a deprivation of due process.” 646 F. 2d 1122, 1124 (1980). As to this question, the panel accurately concluded that “Wainwright v. Sykes, supra, is not applicable to . . . [Isaac’s] petition.” Id.., at 1127.
The full text of Isaac’s claim appears supra, at 139. It is plain that the Court’s claims of “complexity” and “confusion” are merely a smokescreen, behind which the Court feels free to reshape Isaac’s claim.
It does bear some resemblance to Isaac’s claim as construed by the plurality opinion of the Court of Appeals en banc below. 646 F. 2d, at 1133-1136. But the plurality’s construction was simply incorrect, and this Court should correct such errors, not perpetuate them.
433 U. S., at 103-104, and n. 5:
“Under the regime of collateral review recognized since the days of Brown v. Allen [344 U. S. 443 (1953)], and enforced by the Fay bypass test, no rational lawyer would risk the ‘sandbagging’ feared by the Court.5
“5 In brief, the defense lawyer would face two options: (1) He could elect to present his constitutional claims to the state courts in a proper fashion. If the state trial court is persuaded that a constitutional breach has occurred, the remedies dictated by the Constitution would be imposed, the defense would be bolstered, and the prosecution accordingly weakened, *147perhaps precluded altogether. If the state court rejects the properly tendered claims, the defense has lost nothing: Appellate review before the state courts and federal habeas consideration are preserved. (2) He could elect to ‘sandbag.’ This presumably means, first, that he would hold back the presentation of his constitutional claim to the trial court, thereby increasing the likelihood of a conviction since the prosecution would be able to present evidence that, while arguably constitutionally deficient, may be highly prejudicial to the defense. Second, he would thereby have forfeited all state review and remedies with respect to these claims (subject to whatever ‘plain error’ rule is available). Third, to carry out his scheme, he would now be compelled to deceive the federal habeas court and to convince the judge that he did not ‘deliberately bypass’ the state procedures. If he loses on this gamble, all federal review would be barred, and his ‘sandbagging’ would have resulted in nothing but the forfeiture of all judicial review of his client’s claims. The Court, without substantiation, apparently believes that a meaningful number of lawyers are induced into option 2 by Fay. I do not. That belief simply offends common sense.”
McCulloch v. Maryland, 4 Wheat. 316, 407, 415 (1819).
We later relied on Ivan V. in holding that our decision in Mullaney v. Wilbur, 421 U. S. 684 (1975), must be applied retroactively. Hankerson v. North Carolina, 432 U. S. 233, 242-244 (1977).