Opinion of the Court U.S., at 244. Indeed, in Lemon itself, the entanglement that the Court found "independently" to necessitate the program's invalidation also was found to have the effect of inhibiting religion. See, e. g., 403 U. S., at 620 (“[W]e cannot ignore here the danger that pervasive modern governmental power will ultimately intrude on religion . . ."). Thus, it is simplest to recognize why entanglement is significant and treat it as we did in Walz-as an aspect of the inquiry into a statute's effect. Not all entanglements, of course, have the effect of advancing or inhibiting religion. Interaction between church and state is inevitable, see 403 U. S., at 614, and we have always tolerated some level of involvement between the two. Entanglement must be "excessive" before it runs afoul of the Establishment Clause. See, e. g., Bowen v. Kendrick, 487 U. S., at 615-617 (no excessive entanglement where government reviews the adolescent counseling program set up by the religious institutions that are grantees, reviews the materials used by such grantees, and monitors the program by periodic visits); Roemer v. Board of Public Works of Md., 426 U.S. 736, 764-765 (1976) (no excessive entanglement where State conducts annual audits to ensure that categorical state grants to religious colleges are not used to teach religion). The pre-Aguilar Title I program does not result in an "excessive" entanglement that advances or inhibits religion. As discussed previously, the Court's finding of "excessive" entanglement in Aguilar rested on three grounds: (i) the program would require "pervasive monitoring by public authorities" to ensure that Title I employees did not inculcate religion; (ii) the program required “administrative cooperation" between the Board and parochial schools; and (iii) the program might increase the dangers of "political divisiveness." 473 U. S., at 413-414. Under our current understanding of the Establishment Clause, the last two considerations are insufficient by themselves to create an "excessive" Opinion of the Court entanglement. They are present no matter where Title I services are offered, and no court has held that Title I services cannot be offered off campus. Aguilar, supra (limiting holding to on-premises services); Walker v. San Francisco Unified School Dist., 46 F. 3d 1449 (CA9 1995) (same); Pulido v. Cavazos, 934 F. 2d 912, 919-920 (CA8 1991); Committee for Public Ed. & Religious Liberty v. Secretary, United States Dept. of Ed., 942 F. Supp. 842 (EDNY 1996) (same). Further, the assumption underlying the first consideration has been undermined. In Aguilar, the Court presumed that full-time public employees on parochial school grounds would be tempted to inculcate religion, despite the ethical standards they were required to uphold. Because of this risk pervasive monitoring would be required. But after Zobrest we no longer presume that public employees will inculcate religion simply because they happen to be in a sectarian environment. Since we have abandoned the assumption that properly instructed public employees will fail to discharge their duties faithfully, we must also discard the assumption that pervasive monitoring of Title I teachers is required. There is no suggestion in the record before us that unannounced monthly visits of public supervisors are insufficient to prevent or to detect inculcation of religion by public employees. Moreover, we have not found excessive entanglement in cases in which States imposed far more onerous burdens on religious institutions than the monitoring system at issue here. See Bowen, supra, at 615–617. To summarize, New York City's Title I program does not run afoul of any of three primary criteria we currently use to evaluate whether government aid has the effect of advancing religion: It does not result in governmental indoctrination; define its recipients by reference to religion; or create an excessive entanglement. We therefore hold that a federally funded program providing supplemental, remedial instruction to disadvantaged children on a neutral basis is not invalid under the Establishment Clause when such instruc Opinion of the Court tion is given on the premises of sectarian schools by government employees pursuant to a program containing safeguards such as those present here. The same considerations that justify this holding require us to conclude that this carefully constrained program also cannot reasonably be viewed as an endorsement of religion. Accord, Witters, 474 U. S., at 488-489 (“[T]he mere circumstance that [an aid recipient] has chosen to use neutrally available state aid to help pay for [a] religious education [does not] confer any message of state endorsement of religion"); Bowen, supra, at 613-614 (finding no "symbolic link'" when Congress made federal funds neutrally available for adolescent counseling). Accordingly, we must acknowledge that Aguilar, as well as the portion of Ball addressing Grand Rapids' Shared Time program, are no longer good law. C The doctrine of stare decisis does not preclude us from recognizing the change in our law and overruling Aguilar and those portions of Ball inconsistent with our more recent decisions. As we have often noted, "[s]tare decisis is not an inexorable command," Payne v. Tennessee, 501 U. S. 808, 828 (1991), but instead reflects a policy judgment that "in most matters it is more important that the applicable rule of law be settled than that it be settled right,” Burnet v. Coronado Oil & Gas Co., 285 U. S. 393, 406 (1932) (Brandeis, J., dissenting). That policy is at its weakest when we interpret the Constitution because our interpretation can be altered only by constitutional amendment or by overruling our prior decisions. Seminole Tribe of Fla. v. Florida, 517 U. S. 44, 63 (1996); Payne, supra, at 828; St. Joseph Stock Yards Co. v. United States, 298 U. S. 38, 94 (1936) (Stone and Cardozo, JJ., concurring in result) ("The doctrine of stare decisis... has only a limited application in the field of constitutional law"). Thus, we have held in several cases that stare decisis does not prevent us from overruling a previous decision where Opinion of the Court there has been a significant change in, or subsequent development of, our constitutional law. United States v. Gaudin, 515 U. S. 506, 521 (1995) (stare decisis may yield where a prior decision's "underpinnings [have been] eroded, by subsequent decisions of this Court"); Alabama v. Smith, 490 U. S. 794, 803 (1989) (noting that a "later development of . . . constitutional law" is a basis for overruling a decision); Planned Parenthood of Southeastern Pa. v. Casey, 505 U. S. 833, 857 (1992) (observing that a decision is properly overruled where "development of constitutional law since the case was decided has implicitly or explicitly left [it] behind as a mere survivor of obsolete constitutional thinking"). As discussed above, our Establishment Clause jurisprudence has changed significantly since we decided Ball and Aguilar, so our decision to overturn those cases rests on far more than "a present doctrinal disposition to come out differently from the Court of [1985]." Casey, supra, at 864. We therefore overrule Ball and Aguilar to the extent those decisions are inconsistent with our current understanding of the Establishment Clause. Nor does the "law of the case" doctrine place any additional constraints on our ability to overturn Aguilar. Under this doctrine, a court should not reopen issues decided in earlier stages of the same litigation. Messenger v. Anderson, 225 U. S. 436, 444 (1912). The doctrine does not apply if the court is "convinced that [its prior decision] is clearly erroneous and would work a manifest injustice." Arizona v. California, 460 U. S. 605, 618, n. 8 (1983). In light of our conclusion that Aguilar would be decided differently under our current Establishment Clause law, we think adherence to that decision would undoubtedly work a "manifest injustice," such that the law of the case doctrine does not apply. Accord, Davis v. United States, 417 U. S. 333, 342 (1974) (Court of Appeals erred in adhering to law of the case doctrine despite intervening Supreme Court precedent). Opinion of the Court IV We therefore conclude that our Establishment Clause law has "significant[ly] change[d]" since we decided Aguilar. See Rufo, 502 U. S., at 384. We are only left to decide whether this change in law entitles petitioners to relief under Rule 60(b)(5). We conclude that it does. Our general practice is to apply the rule of law we announce in a case to the parties before us. Rodriguez de Quijas v. Shearson/ American Express, Inc., 490 U. S. 477, 485 (1989) (“The general rule of long standing is that the law announced in the Court's decision controls the case at bar"). We adhere to this practice even when we overrule a case. In Adarand Constructors, Inc. v. Peña, 515 U. S. 200 (1995), for example, the District Court and Court of Appeals rejected the argument that racial classifications in federal programs should be evaluated under strict scrutiny, relying upon our decision in Metro Broadcasting, Inc. v. FCC, 497 U. S. 547 (1990). When we granted certiorari and overruled Metro Broadcasting, we did not hesitate to vacate the judgments of the lower courts. In doing so, we necessarily concluded that those courts relied on a legal principle that had not withstood the test of time. 515 U. S., at 237-238. See also Hubbard v. United States, 514 U. S. 695, 715 (1995) (overruling decision relied upon by Court of Appeals and reversing the lower court's judgment that relied upon the overruled case). We do not acknowledge, and we do not hold, that other courts should conclude our more recent cases have, by implication, overruled an earlier precedent. We reaffirm that "[i]f a precedent of this Court has direct application in a case, yet appears to rest on reasons rejected in some other line of decisions, the Court of Appeals should follow the case which directly controls, leaving to this Court the prerogative of overruling its own decisions." Rodriguez de Quijas, supra, at 484. Adherence to this teaching by the District Court and Court of Appeals in this litigation does not insulate a legal principle on which they relied from our review to deter |