concurring in the judgment.
I do not enter the debate about the reasons that took us to the point where mixed constitutional questions are subject to de novo review in federal habeas corpus proceedings. Whatever the answer to that difficult historical inquiry, all agree that, at least prior to the Court’s adoption of the retro-activity analysis of Teague v. Lane, 489 U. S. 288 (1989), see Penry v. Lynaugh, 492 U. S. 302, 313-314 (1989), the matter was settled. It seems that the real issue dividing my colleagues is whether the retroactivity analysis of Teague casts doubt upon the rule of Miller v. Fenton, 474 U. S. 104, 112 (1985). Even petitioner State of Virginia and the United States as amicus curiae, both seeking a deferential standard with respect to mixed questions, recognize that this is how the standard of review question arises. See Brief for Petitioners 11 (“The notion that a state prisoner has a right to de novo federal collateral review of his constitutional claims ... surely has not survived this Court’s decisions in Teague” and its progeny); Brief for United States as Amicus Curiae 12 (“Prior to the rule established by Teague [and later eases applying Teague], this Court often treated mixed questions of law and fact as subject to independent review in federal habeas corpus”).
If vindication of the principles underlying Teague did require that state-court rulings on mixed questions must be given deference in a federal habeas proceeding, then indeed it might be said that the Teague line of cases is on a collision course with the Miller v. Fenton line. And in the proper case we would have to select one at the expense of the other. But in my view neither the purpose for which Teague was *307adopted nor the necessary means for implementing its holding creates any real conflict with the requirement of de novo review of mixed questions.
In my view, it would be a misreading of Teague to interpret it as resting on the necessity to defer to state-court determinations. Teague did not establish a deferential standard of review of state-court decisions of federal law. It established instead a principle of retroactivity. See Teague v. Lane, supra, at 310 (“[W]e now adopt Justice Harlan’s view of retroactivity for cases on collateral review”). To be sure, the fact that our standard for distinguishing old rules from new ones turns on the reasonableness of a state court’s interpretation of then existing precedents suggests that federal courts do in one sense defer to state-court determinations. But we should not lose sight of the purpose of the reasonableness inquiry where a Teague issue is raised: The purpose is to determine whether application of a new rule would upset a conviction that was obtained in accordance with the constitutional interpretations existing at the time of the* prisoner’s conviction.
As we explained earlier this Term:
“When a petitioner seeks federal habeas relief based upon a principle announced after a final judgment, Teague and our subsequent decisions interpreting it require a federal court to answer an initial question, and in some cases a second. First, it must be determined whether the decision relied upon announced a new rule. If the answer is yes and neither exception applies, the decision is not available to the petitioner. If, however, the decision did not announce a new rule, it is necessary to inquire whether granting the relief sought would create a new rule because the prior decision is applied in a novel setting, thereby extending the precedent. The interests in finality, predictability, and comity underlying our new rule jurisprudence may be undermined to an equal degree by the invocation of a rule that was not *308dictated by precedent as by the application of an old rule in a manner that was not dictated by precedent.” Stringer v. Black, 503 U. S. 222, 227-228 (1992) (citation omitted).
The comity interest is not, however, in saying that since the question is close the state-court decision ought to be deemed correct because we are in no better position to judge. That would be the real thrust of a principle based on deference. We see that principle at work in the statutory requirement that, except in limited circumstances, the federal habeas court must presume the correctness of state-court factual findings. See 28 U. S. C. § 2254(d). See also Rushen v. Spain, 464 U. S. 114, 120 (1983) (per curiam) (noting that “the state courts were in a far better position than the federal courts to answer” a factual question). Deference of this kind may be termed a comity interest, but it is not the comity interest that' underlies Teague. The comity interest served by Teague is in not subjecting the States to a regime in which finality is undermined by our changing a rule once thought correct but now understood to be deficient on its own terms. It is in recognition of this principle that we ask whether the decision in question was dictated by precedent. See, e. g., Saffle v. Parks, 494 U. S. 484, 488 (1990).
Teague does bear on applications of law to fact which result in the announcement of a new rule. Whether the prisoner seeks the application of an old rule in a novel setting, see Stringer, supra, at 228, depends in large part on the nature of the rule. If the rule in question is one which of necessity requires a case-by-case examination of the evidence, then we can tolerate a number of specific applications without saying that those applications themselves create a new rule. The rule of Jackson v. Virginia, 443 U. S. 307 (1979), is an example. By its very terms it provides a general standard which calls for some examination of the facts. The standard is whether any rational trier of fact could have found guilt beyond a reasonable doubt after a review of all *309the evidence, so of course there will be variations from case to case. Where the beginning point is a rule of this general application, a rule designed for the specific purpose of evaluating a myriad of factual contexts, it will be the infrequent case that yields a result so novel that it forges a new rule, one not dictated by precedent.
Although as a general matter “new rules will not be applied or announced” in habeas proceedings, Penry, 492 U. S., at 313, there is no requirement that we engage in the threshold Teague inquiry in a case in which it is clear that the prisoner would not be entitled to the relief he seeks even if his case were pending on direct review. See Collins v. Youngblood, 497 U. S. 37 (1990). Therefore, it is not necessary to the resolution of this case to consider the oddity that reversing respondent’s conviction because of the quite fact-specific determination that there was insufficient evidence would have the arguable effect of undercutting the well-established general principle in Virginia and elsewhere that the trier of fact may infer theft from unexplained or falsely denied possession of recently stolen goods. Whether a holding that there was insufficient evidence would constitute one of those unusual cases in which an application of Jackson would create a new rule need not be addressed.
On these premises, the existence of Teague provides added justification for retaining de novo review, not a reason to abandon it. Teague gives substantial assurance that habeas proceedings will not use a new rule to upset a state conviction that conformed to rules then existing. With this safeguard in place, recognizing the importance of finality, de novo review can be exercised within its proper sphere.
For the foregoing reasons, I would not interpret Teague as calling into question the settled principle that mixed questions are subject to plenary review on federal habeas corpus. And, for the reasons I have mentioned, I do not think it necessary to consider whether the respondent brings one of those unusual Jackson claims which is Teague-barred.
*310I agree that the evidence in this case was sufficient to convince a rational factfinder of guilt beyond a reasonable doubt; and I concur in the judgment of the Court.'