Under the Social Security Act, the unmarried minor “child” of a deceased individual who was insured under the Act may receive survivors’ benefits if she was “dependent upon such individual” prior to his death. 49 Stat. 623, as amended, 42 U. S. C. § 402(d)(1)(C) (1988 ed.). In order to determine whether a claimant is, for these purposes, the “child” of the deceased, and, as such, eligible to receive benefits, the Commissioner of Social Security “shall apply such law as would be applied in determining the devolution of intestate personal property by the courts of the State in which [the] insured individual. . . was domiciled at the time of his death.” 42 U. S. C. § 416(h)(2)(A) (1988 ed.).
The petitioner in this case, Lawrence, asserts an entitlement to benefits under these provisions. In so doing, she acknowledges that the relevant state law, that of North Car*165olina, appears on its face to defeat her claim by imposing procedural requirements on proof of paternity (which it requires as a prerequisite for intestate succession) that she cannot meet. She contends, however, that these difficulties can be overcome in her case as they were in Handley v. Schweiker, 697 F. 2d 999 (1983). In that case, the Court of Appeals for the Eleventh Circuit held that state-law requirements of proof of paternity can only be applied against a claimant for benefits under § 416(h)(2)(A) insofar as they are constitutional, and that an Alabama law similar to the North Carolina law involved here was unconstitutional. In contrast, in the case before us, the Court of Appeals for the Fourth Circuit upheld the Social Security Administration’s Appeals Council’s denial of benefits to Lawrence. The Court of Appeals expressly adopted the rationale for rejecting her claim that the Government advanced in its brief to that court: that the constitutionality of a state paternity law need not be considered before applying it to determine entitlement to benefits under the federal statutory scheme. Lawrence petitioned for certiorari to review that decision.
In his response, the Solicitor General advises us that the “Social Security Administration has re-examined” the role of state paternity and intestacy laws in the federal benefits scheme, and now interprets the Social Security Act as “requiring] a determination, at least in some circumstances, of whether the state intestacy statute is constitutional.” Brief for Respondent 8. He also correctly notes that the Act directs the Commissioner of Social Security — not, in the first instance, the courts — to “apply such law as would be applied . . . by the courts of the State” concerned. § 416(h)(2)(A). Without conceding Lawrence’s ultimate entitlement to benefits, he invites us to grant certiorari, vacate the judgment below, and remand the case (GVR) so that the Court of Appeals may either decide it in light of the Commissioner’s new statutory interpretation or remand the case to the Commissioner for reconsideration in light of that interpretation. *166We conclude both that we have the power to issue a GVR order, and that such an order is an appropriate exercise of our discretionary certiorari jurisdiction.
Title 28 U. S. C. § 2106 appears on its face to confer upon this Court a broad power to GVR: “The Supreme Court or any other court of appellate jurisdiction may . . . vacate . . . any judgment, decree, or order of a court lawfully brought before it for review, and may remand the cause and ... require such further proceedings to be had as may be just under the circumstances.” In his dissent issued today in this case and in Stutson v. United States, post, p. 193, another case in which we issue a GVR order, Justice Scalia contends that “traditional practice” and “the Constitution and laws of the United States” impose “implicit limitations” on this power. Post, at 178. We respectfully disagree. We perceive no textual basis for such limitations. The Constitution limits our “appellate Jurisdiction” to issues of “[federal] Law and Fact,” see Art. Ill, § 2, but leaves to Congress the power to “ordain and establish ... inferior Courts,” Art. Ill, § 1, and to make “Exceptions” and “Regulations” limiting and controlling our appellate jurisdiction. Insofar as Congress appears to have authorized such action, we believe that this Court has the power to remand to a lower federal court any case raising a federal issue that is properly before us in our appellate capacity.
Our past practice affirms this conclusion. Although, as Justice Scalia’s dissent explains, post, at 179-183, the exercise of our GVR power was, until recent times, rare, its infrequent early use may be explained in large part by the smaller size of our certiorari docket in earlier times. Regardless of its earlier history, however, the GVR order has, over the past 50 years, become an integral part of this Court’s practice, accepted and employed by all sitting and recent Justices. We have GVR’d in light of a wide range of developments, including our own decisions, see post, at 180 (Scalia, J., dissenting), State Supreme Court decisions, see, *167e. g., Conner v. Simler, 367 U. S. 486 (1961), new federal statutes, see, e. g., Sioux Tribe of Indians v. United States, 329 U. S. 685 (1946), administrative reinterpretations of federal statutes, see, e. g., Schmidt v. Espy, 513 U. S. 801 (1994), new state statutes, see, e. g., Louisiana v. Hays, 512 U. S. 1230 (1994), changed factual circumstances, see, e.g., NLRB v. Federal Motor Truck Co., 325 U. S. 838 (1945) (demilitarization of employees), and confessions of error or other positions newly taken by the Solicitor General, see, e.g., Wells v. United States, 511 U. S. 1050 (1994); Reed v. United States, 510 U. S. 1188 (1994); Ramirez v. United States, 510 U. S. 1103 (1994); Chappell v. United States, 494 U. S. 1075 (1990); Polsky v. Wetherill, 403 U. S. 916 (1971), and state attorneys general, see, e. g., Cuffle v. Avenenti, 498 U. S. 996 (1990); Nicholson v. Boles, 375 U. S. 25 (1963).
This practice has some virtues. In an appropriate case, a GVR order conserves the scarce resources of this Court that might otherwise be expended on plenary consideration, assists the court below by flagging a particular issue that it does not appear to have fully considered, assists this Court by procuring the benefit of the lower court’s insight before we rule on the merits, and alleviates the “[potential for unequal treatment” that is inherent in our inability to grant plenary review of all pending cases raising similar issues, see United States v. Johnson, 457 U. S. 537, 556, n. 16 (1982); cf. Griffith v. Kentucky, 479 U. S. 314, 323 (1987) (“[W]e fulfill our judicial responsibility by instructing the lower courts to apply the new rule retroactively to cases not yet final”). Where intervening developments, or recent developments that we have reason to believe the court below did not fully consider, reveal a reasonable probability that the decision below rests upon a premise that the lower court would reject if given the opportunity for further consideration, and where it appears that such a redetermination may determine the ultimate outcome of the litigation, a GVR order is, we believe, potentially appropriate. Whether a GVR order is ulti*168mately appropriate depends further on the equities of the case: If it appears that the intervening development, such as a confession of error in some, but not all, aspects of the decision below, is part of an unfair or manipulative litigation strategy, or if the delay and further cost entailed in a remand are not justified by the potential benefits of further consideration by the lower court, a GVR order is inappropriate. This approach is similar in its flexibility to this Court’s longstanding approach to applications for stays and other summary remedies granted without determining the merits of the case under the All Writs Act, 28 U. S. C. §1651. See, e. g., Heckler v. Lopez, 463 U. S. 1328 (1983) (Rehnquist, J., in chambers) (staying a District Court order pending the decision on the merits of the Court of Appeals). (Naturally, because GVR orders are premised on matters that we have reason to believe the court below did not fully consider, and because they require only further consideration, the standard that we apply in deciding whether to GVR is somewhat more liberal than the All Writs Act standard, under which relief is granted only upon a showing that a grant of certio-rari and eventual reversal are probable, see id., at 1330.) Used in accordance with this approach, the GVR order can improve the fairness and accuracy of judicial outcomes while at the same time serving as a cautious and deferential alternative to summary reversal in cases whose precedential significance does not merit our plenary review.
Justice Scalia’s dissent would confine GVR’s to three categories of cases:
“(1) where an intervening factor has arisen that has a legal bearing upon the decision, (2) where, in a context not governed by Michigan v. Long, 463 U. S. 1032 (1983), clarification of the opinion below is needed to assure our jurisdiction, and (3) . . . where the respondent or appel-lee confesses error in the judgment below.” Post, at 191-192.
*169While a large proportion of this Court’s GVR orders fall within these categories, we find that, especially as the dissent construes them, they are too narrow to account for the full extent of our actual practice. We find two aspects in particular of the dissent’s approach too restrictive. First, the dissent would insist that only matters that the lower court had no “opportunity” to consider can be the basis for GVR orders. Second, it would impose special restrictions as to when the Court may GVR in light of changes of position by litigants.
Justice Scalia’s dissent concedes — correctly, we believe — that its first category — “intervening factor[s]”— must be extended to include at least Supreme Court decisions rendered so shortly before the lower court’s decision that the lower court had no “opportunity” to apply them. Post, at 181. The dissent does not explain, however, why what the lower court had an “opportunity” to consider should be decisive, or how its “opportunity” is to be assessed. In Robinson v. Story, 469 U. S. 1081 (1984), we GVR’d for further consideration in light of a Supreme Court decision rendered almost three months before the summary affirmance by the Court of Appeals that was the subject of the petition for certiorari. Were those three months sufficient “opportunity” for the court to apprise itself (or be apprised by the parties) of the new, potentially relevant Supreme Court decision? If Robinson was properly GVR’d, we have difficulty understanding the dissent’s objection to our GVR order today in Stutson, where, as in Robinson, the Court of Appeals wrote no opinion to show whether or how it considered a precedent of ours that the District Court had had no opportunity to consider. In both cases, the Court of Appeals “might (or might not) have relied on a standard [nonapplication of the prior Supreme Court decision] that might (or might not) be wrong [and] that might (or might not) have affected the outcome.” Post, at 185 (Scalia, J., dissenting). The only pertinent difference that we can discern between *170the two eases is that the recent Supreme Court precedent was briefed to the Court of Appeals in Stutson, but we have never held lower court briefing to bar our review and vacatur where the lower court’s order shows no sign of having applied the precedents that were briefed. Compare post, at 185-186 (asserting that “we have no power” to vacate and remand in Stutson after relevant briefing and a summary-order below), with, e. g., Netherland v. Tuggle, 515 U. S. 951 (1995) (per curiam) (vacating Court of Appeals’ summary order staying execution for probable failure to apply a 12-year-old Supreme Court precedent that the parties briefed to the Court of Appeals; this Court itself granted a stay a week later, applying that precedent, see Tuggle v. Netherland, 515 U. S. 1188 (1995)). As the prevalence of summary dispositions by the Courts of Appeals continues to increase with the burgeoning federal docket — in 1994, over 11% of Court of Appeals decisions on the merits,1 and many more procedural decisions, were summary — such cases will, no doubt, arise more frequently. In this context, it is important that the meaningful exercise of this Court’s appellate powers not be precluded by uncertainty as to what the court below “might. . . have relied on.” And we are well aware, as are Supreme Court practitioners and lower courts, that while not immune from our plenary review, ambiguous summary dispositions below tend, by their very nature, to lack the precedential significance that we generally look for in deciding whether to exercise our discretion to grant plenary review. We are therefore more ready than the dissent to issue a GVR order in cases in which recent events have cast substantial doubt on the correctness of the lower court’s summary disposition.
With regard to confessions of error and other changes of position by litigants, we agree on several points. All Mem*171bers of the Court are agreed that we “should [not] mechanically accept any suggestion from the Solicitor General that a decision rendered in favor of the Government by a United States Court of Appeals was in error,” Mariscal v. United States, 449 U. S. 405, 406 (1981) (Rehnquist, J., dissenting). And the dissent acknowledges as “well entrenched,” post, at 183 (opinion of Scalia, J.), our practice of GVR’ing in light of plausible confessions of error without determining their merits. Moreover, the dissent is ready in principle to GVR in light of a new agency interpretation of a statute that is entitled to deference under the rule of Chevron U S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837 (1984). Post, at 186-187.
In other respects, however, our approaches to changes of position by litigants diverge. Justice Scalia’s dissent disapproves (although it acknowledges) this Court’s well-established practice of GVR’ing based on confessions of error that do not purport to concede the whole case. Post, at 183, 184-185; cf., e. g., Moore v. United States, 429 U. S. 20 (1976) (GVR’ing based on the Solicitor General’s confession of error, notwithstanding the Solicitor General’s unresolved claim that the error was harmless). The dissent would apparently insist that such GVR’s be confined to cases in which the confession of error concerns a “legal point on which the lower court explicitly relied,” or on which we otherwise “know” for certain that the lower court’s judgment rested. Post, at 185. But, given the legitimacy of GVR’s on the basis of confessions of error without determining the merits, we do not understand why a reasonable probability that the lower court relied on the point at issue should not suffice. As we have explained, supra, at 167, we have GVR’d on the basis of a reasonable probability of a change in result in nonconfession of error cases, see, e. g., Robinson v. Story, supra. We see no special reason why, in a confession of error case, a certainty that the lower court relied on the point in question should be necessary before we may GVR on the basis of a *172reasonable probability that giving the lower court the opportunity to consider that point anew will alter the result.
Similarly, we reject Justice Scalia’s dissent’s other requirement of certainty for GVR’s founded on a change of position by the Government. The dissent accepts in principle that a new interpretation of a statute adopted by the agency charged with implementing it may be entitled to deference in the context of litigation to which the Government is a party. But the dissent would require that before such new interpretation may be the basis for a GVR order, we must be “certain that the change in position is legally cognizable,” post, at 187 (emphasis added), in the sense that it is “entitled to deference,” post, at 188, despite its timing, in that particular case. This requirement, too, appears to be confined to cases in which the event on which the GVR is based is a change of position by the Government, see post, at 187; we do not, for example, understand the dissent to contend that a similar requirement of “lega[l] cognizab[ility]” should apply to GVR’s in habeas corpus cases in which the procedural bar that we recognized in Teague v. Lane, 489 U. S. 288 (1989), might apply. Again, we do not understand the rationale for imposing such special requirements on GVR’s based on a change of position. If it appears reasonably probable that a confession of error reveals a genuine and potentially determinative error by the court below, a GVR may be appropriate; similarly, we believe that if an agency interpretation is reasonably probably entitled to deference and potentially determinative, we may GVR in light of it. It is precisely because of uncertainty that we GVR. We do not see why uncertainty as to the “lega[l] cognizab[ility]” of an agency interpretation in a particular case should be treated differently from uncertainty as to its application in that case. Indeed, to determine on the merits whether deference is owed to the agency interpretation, based on a circumstance — i. e., its timing with respect to the case at hand — that will not be present in any other case brought *173under the statute at issue, when the Court of Appeals has had no “opportunity” to consider the new agency interpretation, appears to us to defeat the purpose of GVR’ing. Rather, we think it appropriate to apply our normal “reasonable probability” test, and to defer any special concerns about strategic litigating behavior that are raised by changes in the Government’s position to consideration of the equities. Under our approach, neither uncertainty as to whether the Government’s change of position, if accepted, would be outcome determinative, nor uncertainty as to the “lega[l] cog-nizab[ility]” of an administrative interpretation, preclude a GVR if the overall probabilities and equities support the GVR order. Indeed, we issued just such a GVR order last Term, without recorded dissent. See Schmidt v. Espy, 513 U. S. 801 (1994) (GVR on the basis of the Solicitor General’s representation that “[ajfter further examination of the regulation and its application in the present case,... the Department of Agriculture has determined that petitioners’ lease-baek/buybaek application should be reconsidered without respect to the good faith requirement,” Brief for Respondent in Schmidt v. Espy, O. T. 1994, No. 93-1707, p. 7; see also id., at 10, n. 5 (maintaining that other obstacles to petitioners’ application might remain)).
Our differences with Justice Scalia’s dissent should not overshadow the substantial level of agreement shared by all Members of this Court. On the one hand, all are agreed that a wide range of intervening developments, including confessions of error, may justify a GVR order. On the other hand, all are agreed that our GVR power should be exercised sparingly. This Court should not just GVR a case “because it finds the opinion, though arguably correct, incomplete and unworkmanlike; or because it observes that there has been a postjudgment change in the personnel of the state supreme court, and wishes to give the new state justices a shot at the case.” Post, at 189 (Scalia, J., dissenting); accord, Alvarado v. United States, 497 U. S. 543, 545 (1990) (Rehnquist, *174C. J., dissenting). Respect for lower courts, the public interest in finality of judgments, and concern about our own expanding certiorari docket all counsel against undisciplined GVR’ing. It remains to apply these principles to the facts of this case.
The feature of this case that, in our view, makes a GVR order appropriate is the new interpretation of the Social Security Act that the Solicitor General informs us that the Social Security Administration, the agency charged with implementing that Act, has adopted. As Justice Scalia’s dissent notes, post, at 187, we have not settled whether and to what extent deference is due to an administrative interpre-ation — its “lega[l] cognizab[ility]” — in a case that has already reached the appeal or certiorari stage when that interpretation is adopted. But in our view, see supra, at 172-173, such uncertainty does not preclude a GVR. Indeed, it is precisely because we are uncertain, without undertaking plenary analysis, of the legal impact of a new development, especially one, such as the present, which the lower court has had no opportunity to consider, that we GVR. Here, as in Schmidt, supra, the Solicitor General has recommended judicial reconsideration of the merits, while not conceding the petitioner’s ultimate entitlement to statutory benefits, based on a new statutory interpretation that will apparently be applied, and will probably be entitled to deference, in future cases nationwide. Here, as in Schmidt, our summary review leads us to the conclusion that there is a reasonable probability that the Court of Appeals would conclude that the timing of the agency’s interpretation does not preclude the deference that it would otherwise receive, and that it may be outcome determinative in this case. A GVR order is, therefore, appropriate, subject to the equities.
As to the equities, it seems clear that they favor a GVR order here. That disposition has the Government’s express support, notwithstanding that its purpose is to give the Court of Appeals the opportunity to consider an administrative interpretation that appears contrary to the Govern*175ment’s narrow self-interest. And the Government has informed us that it intends to apply that interpretation to future cases nationwide. Giving Lawrence a chance to benefit from it furthers fairness by treating Lawrence like other future benefits applicants. We acknowledge the dissent’s concern that postlitigation interpretations may be the product of unfair or manipulative Government litigating strategies, see post, at 187, and we therefore view late changes of position by the Government with some skepticism. That general concern does not, however, appear to us to require that we deprive Lawrence of the benefit of a favorable administrative reinterpretation in these particular circumstances. We believe, therefore, that the equities and legal uncertainties of this case together merit a GVR order.2
Accordingly, the motion for leave to proceed in forma pau-peris and the petition for a writ of certiorari are granted. The judgment is vacated and the case is remanded to the United States Court of Appeals for the Fourth Circuit for further consideration in light of the position taken in the brief for respondent filed by the Solicitor General, August 17,1995.
See Administrative Office of United States Courts, Reports of Proceedings of Judicial Conference of the United States, 1994, table S-3 (3,080 out of 27,219 decisions on the merits).
In a letter filed on October 24,1995, the Solicitor General advised this Court of a July 1995 amendment to the North Carolina paternity statute, N. C. Gen. Stat. §49-14(c). We find it unnecessary to decide whether this development independently justifies our GVR order. The Court of Appeals is free to consider its significance on remand.