with whom Justice Brennan, Justice Marshall, and Justice Stevens join, dissenting.
In 1981, Congress enacted the Adolescent Family Life Act (AFLA), 95 Stat. 578, 42 U. S. C. §300z et seq. (1982 ed. and Supp. IV), thereby “involv[ing] families[,} . . . religious and charitable organizations, voluntary associations, and other groups,” §3Q0z-5(a)(21), in a broad-scale effort to alleviate some of the problems associated with teenage pregnancy. It is unclear whether Congress ever envisioned that public funds would pay for a-program during a session of which parents and teenagers would be instructed:
“You want to know the church teachings on sexuality. . . . You are the church. You people sitting here are the body of Christ. The teachings of you and the things you value are, in fact, the values of the Catholic Church.” App. 226.
Or of curricula that taught:
“The Church has always taught that the marriage act, or intercourse, seals the union of husband and wife, (and is a representation of their union on all levels.) Christ commits Himself to us when we come to ask for the sacrament of marriage. We ask Him to be active in our life. God is love. We ask Him to share His love in ours, and God procreates with us, He enters into our physical union with Him, and we begin new life.” Id., at 372.
Or the teaching of a method of family planning described on the grant application as “not only a method of birth regulation but also a philosophy of procreation,” id., at 143, and promoted as helping “spouses who are striving ... to transform their married life into testimony[,]... to cultivate their matrimonial spirituality!, and] to make themselves better in*626struments in God’s plan,” and as “facilitating] the evange-lization of homes.” Id,., at 385.
Whatever Congress had in mind, however, it enacted a statute that facilitated and, indeed, encouraged the use of public funds for such instruction, by giving religious groups a central pedagogical and counseling role without imposing any restraints on the sectarian quality of the participation. As the record .developed thus far in this litigation makes all too clear, federal tax dollars appropriated for AFLA purposes have been used, with Government approval, to support religious teaching. Today the majority upholds the facial validity of this statute and remands the action to the District Court for further proceedings concerning appellees’ challenge to the manner in which the statute has been applied. Because I am firmly convinced that our cases require invalidating this statutory scheme, I dissent.
I
The District Court, troubled by the lack of express guidance from this Court as to the appropriate manner in which to examine Establishment Clause challenges to an entire statute as well as to specific instances of its implementation, reluctantly proceeded to analyze the AFLA both “on its face” and “as applied.” Thereafter, on cross-motions for summary judgment supported by an extensive record of undisputed facts, the District Court applied the three-pronged analysis of Lemon v. Kurtzman, 403 U. S. 602 (1971), and declared the AFLA unconstitutional both facially and as applied. 657 F. Supp. 1547 (DC 1987). The majority acknowledges that this Court in some cases has passed on the facial validity of a legislative enactment and in others limited its analysis to the particular applications at issue; yet, while confirming that the District Court was justified in analyzing the AFLA both ways, the Court fails to elaborate on the consequences that flow from the analytical division.
*627While the distinction is sometimes useful in constitutional litigation, the majority misuses it here to divide and conquer appellees’ challenge.1 By designating appellees’ broad attack on the statute as a “facial” challenge, the majority justifies divorcing its analysis from the extensive record developed in the District Court, and thereby strips the challenge of much of its force and renders the evaluation of the Lemon “effects” prong particularly sterile and meaningless. By characterizing appellees’ objections to the real-world operation of the AFLA an “as-applied” challenge, the Court risks misdirecting the litigants and the lower courts toward piecemeal litigation continuing indefinitely throughout the life of the AFLA. In my view, a more effective way to review Establishment Clause challenges is to look to the type of re*628lief prayed for by the plaintiffs, and the force of the arguments and supporting evidence they marshal. Whether we denominate a challenge that focuses on the systematically unconstitutional operation of a statute a “facial” challenge — because it goes to the statute as a whole — or an “as-applied” challenge — because we rely on real-world events — the Court should not blind itself to the facts revealed by the undisputed record.2
As is evident from the parties’ arguments, the record compiled below, and the decision of the District Court, this lawsuit has been litigated primarily as a broad challenge to the statutory scheme as a whole, not just to the awarding of grants to a few individual applicants. The thousands of pages of depositions, affidavits, and documentary evidence were not intended to. demonstrate merely that particular grantees should not receive further funding. Indeed, because of the 5-year grant cycle, some of the original grantees are no longer AFLA participants. This record was designed to show that the AFLA had been interpreted and implemented by the Government in a manner that was clearly unconstitutional, and appellees sought declaratory and injunc-tive relief as to the entire statute.
*629In discussing appellees’ as-applied challenge, the District Court recognized that their objections went further than the validity of the particular grants under review:
“The undisputed record before the Court transforms the inherent conflicts between the AFLA and the Constitution into reality. . . . While the Court will not engage in an exhaustive recitation of the record, references to representative portions of the record reveal the extent to which the AFLA has in fact ‘directly and immediately’ advanced religion, funded ‘pervasively sectarian’ institutions, or permitted the use of federal tax dollars for education and counseling that amounts to the teaching of religion.” 657 F. Supp., at 1564 (footnote omitted).
The majority declines to accept the District Court’s characterization of the record, yet fails to review it independently, relying instead on its assumptions and casual observations about the character of the grantees and potential grantees.3 *630See ante, at 610, 611-612, 616-617. In doing so, the Court neglects its responsibilities under the Establishment Clause and gives uncharacteristically short shrift to the District Court’s understanding of the facts.4
II
Before proceeding to apply Lemon’s three-part analysis to the AFLA, I pause to note a particular flaw in the majority’s method. A central premise of the majority opinion seems to be that the primary means of ascertaining whether a statute that appears to be neutral on its face in fact has the effect of advancing religion is to determine whether aid flows to “pervasively sectarian” institutions. See ante, at 609-610, 616, 621. This misplaced focus leads the majority to ignore the substantial body of case law the Court has developed in analyzing programs providing direct aid to parochial schools, *631and to rely almost exclusively on the few cases in which the Court has upheld the supplying of aid to private colleges, including religiously affiliated institutions.
“Pervasively sectarian,” a vaguely defined term of art, has its roots in this Court’s recognition that government must not engage in detailed supervision of the inner workings of religious institutions, and the Court’s sensible distaste for the “picture of state inspectors prowling the halls of parochial schools and auditing classroom instruction,” Lemon v. Kurtzman, 403 U. S., at 650 (Brennan, J., concurring); see also Aguilar v. Felton, 473 U. S. 402, 411 (1985); Roemer v. Maryland Public Works Board, 426 U. S. 736, 762 (1976) (plurality opinion). Under the “effects” prong of the Lemon test, the Court has used one variant or another of the pervasively sectarian concept to explain why any but the most indirect forms of government aid to such institutions would necessarily have the effect of advancing religion. For example, in Meek v. Pittenger, 421 U. S. 349, 365 (1975), the Court explained:
“[I]t would simply ignore reality to attempt to separate secular educational functions from the predominantly religious role performed by many of Pennsylvania’s church-related elementary and secondary schools and to then characterize Act 195 as ■ channeling aid to the secular without providing direct aid to the sectarian.”
See also Hunt v. McNair, 413 U. S. 734, 743 (1973).
The majority first skews the Establishment Clause analysis by adopting a cramped view of what constitutes a pervasively sectarian institution. Perhaps because most of the Court’s decisions in this area have come in the context of aid to parochial schools, which traditionally have been characterized as- pervasively sectarian, the majority seems to equate the characterization with the institution.5 In support of that *632illusion, the majority relies heavily on three cases in which the Court has upheld direct government funding to liberal arts colleges with some religious affiliation, noting that such colleges were not “pervasively sectarian.” But the happenstance that the few'cases in which direct-aid statutes have been upheld have concerned religiously affiliated liberal arts colleges no more suggests that only parochial schools should be considered “pervasively sectarian,” than it suggests that the only religiously affiliated institutions that may ever receive direct government funding are private liberal arts colleges. In fact, the cases on which the majority relies have stressed that the institutions’ “predominant higher education mission is to provide their students with a secular education.” Tilton v. Richardson, 403 U. S. 672, 687 (1971) (emphasis added); see Roemer v. Maryland Public Works Board, 426 U. S., at 755 (noting “high degree of institutional autonomy” and that “the encouragement of spiritual development is only one secondary objective of each college”) (internal quotations omitted); Hunt v. McNair, 413 U. S., at 744 (finding “no basis to conclude that the College’s operations are oriented significantly towards sectarian rather than secular education”). In sharp contrast, the District Court here concluded that AFLA grantees and participants included “organizations with institutional ties to religious denominations and corporate requirements that the organizations abide by and not contradict religious doctrines. In addition, other recipients of AFLA funds, while not explicitly affiliated with a religious denomination, are religiously inspired and dedicated to teaching the dogma that inspired them” (emphasis *633added). 657 F. Supp., at 1564. On a continuum of “sectarianism” running from parochial schools at one end to the colleges funded by the statutes upheld in Tilton, Hunt, and Roemer at the other, the AFLA grantees described by the District Court clearly are much closer to the former than to the latter.
More importantly, the majority also errs in suggesting that the inapplicability of the label is generally dispositive. While a plurality of the Court has framed the inquiry as “whether an institution is so ‘pervasively sectarian’ that it may receive no direct state aid of any kind,” Roemer v. Maryland Public Works Board, 426 U. S., at 758, the Court never has treated the absence of such a finding as a license to disregard the potential for impermissible fostering of religion. The characterization of an institution as “pervasively sectarian” allows us to eschew further inquiry into the use that will be made of direct government aid. In that sense, it is a sufficient, but not a necessary, basis for a finding that a challenged program creates an. unacceptable Establishment Clause risk. The label thus serves in some cases as a proxy for a more detailed analysis of the institution, the nature of the aid, and the manner in which the aid may be used.
The voluminous record compiled by the parties and reviewed by the District Court illustrates the manner in which the AFLA has been interpreted and implemented by the agency responsible for the aid program, and eliminates whatever need there might be to speculate about what kind of institutions might receive funds and how they might be selected; the record explains the nature of the activities funded with Government money, as well as the content of the educational programs and materials developed and disseminated. There is no basis for ignoring the volumes of depositions, pleadings, and undisputed facts reviewed by the District Court simply because the recipients of the Government funds may not in every sense resemble parochial schools.
*634HH h — I H-i
As is often the case, it is the effect of the statute, rather than its purpose, that creates Establishment Clause problems. Because I have no meaningful disagreement with the majority’s discussion of the AFLA’s essentially secular purpose, and because I find the statute’s effect of advancing religion dispositive, I turn to that issue directly.
A
The majority’s holding that the AFLA is not unconstitutional on its face marks a sharp departure from our precedents. While aid programs providing nonmonetary, verifi-ably secular aid have been upheld notwithstanding the indirect effect they might have on the allocation of an institution’s own funds for religious activities, see, e. g., Board of Education v. Allen, 392 U. S. 236 (1968) (lending secular textbooks to parochial schools); Everson v. Board of Education, 330 U. S. 1 (1947) (providing bus services to parochial schools), direct cash subsidies have always required much closer scrutiny into the expected and potential uses of the funds, and much greater guarantees that the funds would not be used inconsistently with the Establishment Clause. Parts of the AFLA prescribing various forms of outreach, education, and counseling services6 specifically authorize the expenditure of funds in ways previously held unconstitutional. For example, the Court has upheld the use of public funds to support a parochial school’s purchase of secular textbooks already approved for use in public schools, see Wolman v. Walter, 433 U. S. 229, 236-238 (1977); Meek v. Pittenger, 421 U. S., at 359-362, or its grading and administering of state-prepared tests, Committee for Public Education & Religious Liberty v. Regan, 444 U. S. 646 (1980). When the books, teaching materials, or examinations were to *635be selected or designed by the private schools themselves, however, the Court consistently has held that such government aid risked advancing religion impermissibly. See, e. g., Wolman v. Walter, 433 U. S., at 248-251; Levitt v. Committee for Public Education & Religious Liberty, 413 U. S. 472 (1973); Lemon v. Kurtzman, 403 U. S., at 620-621. The teaching materials that may be purchased, developed, or disseminated with AFLA funding are in no way restricted to those already selected and approved for use in secular contexts.7
Notwithstanding the fact that Government funds are paying for religious organizations to teach and counsel impressionable adolescents on a highly sensitive subject of considerable religious significance, often on the premises of a church or parochial school and without any effort to remove religious symbols from the sites, 657 F. Supp., at 1565-1566, the majority concludes that the AFLA is not facially invalid. The majority acknowledges the constitutional proscription on *636government-sponsored religious indoctrination but, on the basis of little more than an indefensible assumption that AFLA recipients are not pervasively sectarian and consequently are presumed likely to comply with statutory and constitutional mandates, dismisses as insubstantial the risk that indoctrination will enter counseling. Ante, at 611-612. Similarly, the majority rejects the District Court’s conclusion that the subject matter renders the risk of indoctrination unacceptable, and does so, it says, because “the likelihood that some of the religious institutions who receive AFLA funding will agree with the message that Congress intended to deliver to adolescents through the AFLA” does not amount to the advancement of religion. Ante, at 613. I do not think the statute can be so easily and conveniently saved.
(1)
The District Court concluded that asking religious organizations to teach and counsel youngsters on matters of deep religious significance, yet expect them to refrain from making reference to religion is both foolhardy and unconstitutional. The majority’s rejection of this view is illustrative of its doctrinal misstep in relying so heavily on the college-funding cases. The District Court reasoned:
“To presume that AFLA counselors from religious organizations can put their beliefs aside when counseling an adolescent on matters that are part of religious doctrine is simply unrealistic. . . . Even if it were possible, government would tread impermissibly on religious liberty merely by suggesting that religious organizations instruct on doctrinal matters without any conscious or unconscious reference to that doctrine. Moreover, the statutory scheme is fraught with the possibility that religious beliefs might infuse instruction and never be detected by the impressionable and unlearned adolescent to whom the instruction is directed” (emphasis in original). 657 F. Supp., at 1563.
*637The majority rejects the District Court’s assumptions as unwarranted outside the context of a pervasively sectarian institution. In doing so, the majority places inordinate weight on the nature of the institution receiving the funds, and ignores altogether the targets of the funded message and the nature of its content.
I find it nothing less than remarkable that the majority relies on statements expressing confidence that administrators of religiously affiliated liberal arts colleges would not breach statutory proscriptions and use government funds earmarked “for secular purposes only,” to finance theological instruction or religious worship, see ante, at 612, citing Roemer, 426 U. S., at 760-761, and Tilton, 403 U. S., at 682, in order to reject a challenge based on the risk of indoctrination inherent in “educational services relating to family life and problems associated with- adolescent premarital sexual relations,” or “outreach services to families of adolescents to discourage sexual relations among unemancipated minors.” §§ 300z-l(a)(4)(G), (0). The two situations are simply not comparable.8
*638The AFLA, unlike any statute this Court has upheld, pays for teachers and counselors, employed by and subject to the direction of religious authorities, to educate impressionable young minds on issues of religious moment; Time and again we have recognized the difficulties inherent in asking even the best-intentioned individuals in such positions to make “a total separation between secular teaching and religious doctrine.” Lemon v. Kurtzman, 403 U. S., at 619. Accord, Levitt v. Committee for Public Education & Religious Liberty, 413 U. S., at 481; Meek v. Pittenger, 421 U. S., at 370-371; Roemer v. Maryland Public Works Board, 426 U. S., at 749 (plurality opinion); Wolman v. Walter, 433 U. S., at 254; Grand Rapids School District v. Ball, 473 U. S. 373, 388 (1985). Where the targeted audience is composed of children, of course, the Court’s insistence on adequate safeguards has always been greatest. See, e. g., Grand Rapids School District v. Ball, 473 U. S., at 383, 390; Committee for Public Education & Religious Liberty v. Nyquist, 413 U. S. 756, 796-798 (1973), Lemon v. Kurtzman, 403 U. S., at 622-624. In those cases in which funding of colleges with religious affiliations has been ,upheld, the Court has relied on the assumption that “college students are less impressionable and less susceptible to religious indoctrination. . . . The skepticism of the college student is not an inconsiderable barrier to any attempt or .tendency to subvert the congressional objectives and limitations” (footnote omitted). Tilton v. Richardson, 403 U. S., at 686 (plurality opinion). See also Widmar v. Vincent, 454 U. S. 263, 274, n. 14 (1981) (“University students are, of course, young adults. They are less impressionable than younger students and should be able to appreciate that the University’s policy is one of neutrality toward religion”).
(2)
By observing that the alignment of the statute and the religious views of the grantees do not render the AFLA a statute which funds “specifically religious activity,” the majority *639makes light of the religious significance in the counseling provided by some grantees. Yet this is a dimension that Congress specifically sought to capture by enlisting the aid of religious organizations in battling the problems associated with teenage pregnancy. See S. Rep. No. 97-161, pp. 15-16 (1981); S. Rep. No. 98-496, pp. 9-10 (1984). Whereas there may be secular values promoted by the AFLA, including the encouragement of adoption and premarital chastity and the discouragement of abortion, it can hardly be doubted that when promoted in theological terms by religious figures, those values take on a religious nature. Not surprisingly, the record is replete with observations to that effect.9 It *640should be undeniable by now that religious dogma may not be employed by government even to accomplish laudable secular purposes such as “the promotion of moral values, the contradiction to the materialistic trends of our times, the perpetuation of our institutions and the teaching of literature.” Abington School District v. Schempp, 374 U. S. 203, 223 (1963) (holding unconstitutional daily reading of Bible verses and recitation of the Lord’s Prayer in public schools); Stone v. Graham, 449 U. S. 39 (1980) (holding unconstitutional posting of Ten Commandments despite notation explaining secular application thereof).10
It is true, of course, that the Court has recognized that the Constitution does not prohibit the government from supporting secular social-welfare services solely because they are provided by a religiously affiliated organization. See ante, at 609. But such recognition has been closely tied to the nature of the subsidized social service: “the State may send a *641cleric, indeed even a clerical order, to perform a wholly secular task” (emphasis added). Roemer v. Maryland Public Works Board, 426 U. S., at 746 (plurality opinion). There is a very real and important difference between running a soup kitchen or a hospital, and counseling pregnant teenagers on how to make the difficult decisions facing them. The risk of advancing religion at public expense, and of creating an appearance that the government is endorsing the medium and the message, is much greater when the religious organization is directly engaged in pedagogy, with the express intent of shaping belief and changing behavior, than where it is neutrally dispensing medication, food, or shelter.11
There is also, of course, a fundamental difference between government’s employing religion because of its unique appeal to a higher authority and the transcendental nature of its message, and government’s enlisting the aid of religiously committed individuals or organizations without regard to their sectarian motivation. In the latter circumstance, religion plays little or no role; it merely explains why the individual or organization has chosen to get involved in the publicly funded program. In the former, religion is at the core of the subsidized activity, and it affects the manner in which the “service” is dispensed. For some religious organizations, *642the answer to a teenager’s question “Why shouldn’t I have an abortion?” or “Why shouldn’t I use barrier contraceptives?” will undoubtedly be different from an answer based solely on secular considerations.12 Public funds may not be used to endorse the religious message.
B
The problems inherent in a statutory scheme specifically designed to involve religious organizations in a government-funded pedagogical program are compounded by the lack of any statutory restrictions on the use of federal tax dollars to promote religion. Conscious of the remarkable omission from the AFLA of any restriction whatsoever on the use of public funds for sectarian purposes, the Court disingenuously argues that we have “never stated that a statutory restriction is constitutionally required.” Ante, at 614. In Tilton v. Richardson, this Court upheld a statute providing grants and loans to colleges for the construction of academic facilities because it “expressly prohibited] their use for religious instruction, training, or worship . . . and the record show[ed] that some church-related institutions ha[d] been required to disgorge benefits for failure to obey” the restriction, 403 U. S., at 679-680, but severed and struck a provision of the statute that permitted the restriction to lapse after 20 years. The Tilton Court noted that the statute required applicants to *643provide assurances only that use of the funded facility would be limited to secular purposes for the initial 20-year period, and that this limitation, “obviously opens the facility to use for any purpose at the end of that period.” Id., at 683. Because they expired after 20 years, “the statute’s enforcement provisions [were] inadequate to ensure that the impact of the federal aid will not advance religion.” Id., at 682.
' The majority interprets Tilton “to indicate that the constitutional limitations on use of federal funds, as embodied in the statutory restriction, could not simply ‘expire’ ” after 20 years, but concludes that the absence of a statutory restriction in the AFLA is not troubling, because “there is also no intimation in the statute that at some point, or for some grantees, religious uses are permitted.” Ante, at 614. Although there is something to the notion that the lifting of a pre-existing restriction may be more likely to be perceived as affirmative authorization than would the absence of any restriction at all, there was in Tilton no provision that stated that after 20 years facilities built under the aid program could be converted into chapels. What there was in Tilton was an express statutory provision, which lapsed, leaving no restrictions; it was that vacuum that the Court found constitutionally impermissible. In the AFLA, by way of contrast, there is a vacuum right from the start.13
*644If Tilton were indeed the only indication that cash-grant programs must include prohibitions on the use of public funds to advance or endorse religion, one might argue more plausibly that ordinary reporting requirements, in conjunction with some presumption that Government agencies administer federal programs in a constitutional fashion,14 might suffice to *645protect a statute against facial challenge. That, however, is simply not the case. In Committee for Public Education & Religious Liberty v. Regan, 444 U. S. 646 (1980), for example, the Court upheld a state program whereby private schools were reimbursed for the actual cost of administering state-required tests. The statute specifically required that no payments be made for religious instruction and incorporated an extensive auditing system. The Court warned, however: “Of course, under the relevant cases the outcome would likely be different were there no effective means for insuring that the cash reimbursements would cover only secular services.” Id., at 659. In this regard, the Regan Court merely echoed and reaffirmed what was already well established. In Committee for Public Education & Religious Liberty v. Nyquist, the Court explained:
“Nothing in the statute, for instance, bars a qualifying school from paying out of state funds the salaries of employees who maintain the school chapel, or the cost of renovating classrooms in which religion is taught, or the cost of heating and lighting those same facilities. Absent appropriate restrictions on expenditures for these and similar purposes, it simply cannot be denied that this section has a primary effect that advances religion in that it subsidizes directly the religious activities of sectarian elementary and secondary schools” (emphasis added). 413 U. S., at 774.
*646See id., at 780 (“In the absence of an effective means of guaranteeing that the state aid derived from public funds will be used exclusively for secular, neutral, and nonideological purposes, it is clear from our cases that direct aid in whatever form is invalid”); Lemon v. Kurtzman, 403 U. S., at 621 (“The history of government grants of a continuing cash subsidy indicates that such programs have almost always been accompanied by varying measures of control and surveillance”). See also Roemer v. Maryland Public Works Board, 426 U. S., at 760 (upholding grant program containing statutory restriction on using state funds for “sectarian purposes”); Hunt v. McNair, 413 U. S., at 744 (noting that the statute at issue “specifically states that a project ‘shall not include’ any buildings or facilities used for religious purposes”).15
Despite the glaring omission of a restriction on the use of funds for religious purposes, the Court attempts to resurrect the AFLA by noting a legislative intent not to promote religion, and observing that various reporting provisions of the statute “create a mechanism whereby the Secretary can police the grants.” Ante, at 615. However effective this “mechanism” might prove to be in enforcing clear statutory directives, it is of no help where, as here, no restrictions are found on the face of the statute, and the Secretary has not promulgated any by regulation. Indeed, the only restriction *647on the use of AFLA funds for religious purposes is found in the Secretary’s “Notice of Grant Award” sent to grantees, which specifies that public funds may not be used to “teach or promote religion,” 657 F. Supp., at 1563, n. 13, and apparently even that clause was not inserted until after this litigation was underway. Furthermore, the “enforcement” of the limitation on sectarian use of AFLA funds, such as it is, lacks any bite. There is no procedure pursuant to which funds used to promote religion must be refunded to the Government, as there was, for example, in Tilton v. Richardson, 403 U. S., at 682.
Indeed, nothing in the AFLA precludes the funding of even “pervasively sectarian” organizations, whose work by definition cannot be segregated into religious and secular categories. And, unlike a pre-enforcement challenge, where there is no record to review, or a limited challenge to a specific grant, where the Court is reluctant to invalidate a statute “in anticipation that particular applications may result in unconstitutional use of funds,” Roemer v. Maryland Public Works Board, 426 U. S., at 761, in this litigation the District Court expressly found that funds have gone to pervasively sectarian institutions and tax dollars have been used for the teaching of religion. 657 F. Supp., at 1564. Moreover, ap-pellees have specifically called into question the manner in which the grant program was administered and grantees were selected. See n. 14, supra. These objections cannot responsibly be answered by reliance on the Secretary’s enforcement mechanism. See, e. g., Levitt v. Committee for Public Education & Religious Liberty, 413 U. S., at 480 (“[T]he State is constitutionally compelled to assure that the state-supported activity is not being used for religious indoctrination”); Lemon v. Kurtzman, 403 U. S., at 619 (“The State must be certain, given the Religion Clauses, that subsidized teachers do not inculcate religion”).
*648c
By placing unsupportable weight on the “pervasively sectarian” label, and recharacterizing appellees’ objections to the statute, the Court attempts to create an illusion of consistency between our prior cases and its present ruling that the AFLA is not facially invalid. But the Court ignores the unwavering vigilance that the Constitution requires against any law “respecting an establishment of religion,” U. S. Const., Arndt. 1, which, as we have recognized time and again, calls for fundamentally conservative decisionmaking: our cases do not require a plaintiff to demonstrate that a government action necessarily promotes religion, but simply that it creates such a substantial risk. See, e. g., Grand Rapids School District v. Ball, 473 U. S., at 387 (observing a “substantial risk that, overtly or subtly, the religious message . . . will infuse the supposedly secular classes”); Committee for Public Education & Religious Liberty v. Regan, 444 U. S., at 656 (describing as “minimal” the chance that religious bias would enter process of grading state-drafted tests in secular subjects, given “complete” state safeguards); Wolman v. Walter, 433 U. S., at 254 (noting “unacceptable risk of fostering of religion” as “an inevitable byproduct” of teacher-accompanied field trips); Meek v. Pittenger, 421 U. S., at 372 (finding “potential for impermissible fostering of religion”); Levitt v. Committee for Public Education & Religious Liberty, 413 U. S., at 480 (finding dispositive “the substantial risk that . . . examinations, prepared by teachers under the authority of religious institutions, will be drafted with an eye, unconsciously or otherwise, to inculcate students in the religious precepts of the sponsoring church”); Lemon v. Kurtzman, 403 U. S., at 619 (finding “potential for impermissible fostering of religion”). Given the nature of the subsidized activity, the lack of adequate safeguards, and the chronicle of past experience with this statute, there is no room for doubt that the AFLA creates a substantial risk of impermissible fostering of religion.
*649 <
While it is evident that the AFLA does not pass muster under Lemon’s “effects” prong, the unconstitutionality of the statute becomes even more apparent when we consider the unprecedented degree of entanglement between Church and State required to prevent subsidizing the advancement of religion with AFLA funds. The majority’s brief discussion of Lemon’s “entanglement” prong is limited to (a) criticizing it as a “Catch-22,” and (b) concluding that because there is “no reason to assume that the religious organizations which may receive grants are ‘pervasively sectarian’ in the same sense as the Court has held parochial schools to be,” there is no need to be concerned about the degree of monitoring which will be necessary to ensure compliance with the AFLA and the Establishment Clause. Ante, at 615-616. As to the former, although the majority is certainly correct that the Court’s entanglement analysis has been criticized in the separate writings of some Members of the Court, the question whether a government program leads to “ ‘an excessive government entanglement with religion’ ” nevertheless is and remains a part of the applicable constitutional inquiry. Lemon v. Kurtzman, 403 U. S., at 613, quoting Walz v. Tax Comm’n, 397 U. S. 664, 674 (1970). I accept the majority’s conclusion that “[tjhere is no doubt that the monitoring of AFLA grants is necessary ... to ensure that public money is to be spent ... in a way that comports with the Establishment Clause,” ante, at 615, but disagree with its easy characterization of entanglement analysis as a “Catch-22.” To the extent any metaphor is helpful, I would be more inclined to characterize the Court’s excessive entanglement decisions as concluding that to implement the required monitoring, we would have to kill the patient to cure what ailed him. See, e. g., Lemon v. Kurtzman, 403 U. S., at 614-615; Meek v. Pittenger, 421 U. S., at 370; Aguilar v. Felton, 473 U. S., at 413-414.
*650As to the Court’s conclusion that our precedents do not indicate that the Secretary’s monitoring will have to be exceedingly intensive or entangling, because the grant recipients are not sufficiently like parochial schools, I must disagree. As discussed above, the majority’s excessive reliance on the distinction between the Court’s parochial-school-aid cases and college-funding cases is unwarranted. Lemon, Meek, and Aguilar cannot be so conveniently dismissed solely because the majority declines to assume that the “pervasively sectarian” label can be applied here.
To determine whether a statute fosters excessive entanglement, a court must look at three factors: (1) the character and purpose of the institutions benefited; (2) the nature of the aid; and (3) the nature of the relationship between the government and the religious organization. See Lemon v. Kurtzman, 403 U. S., at 614-615. Thus, in Lemon, it was not solely the fact that teachers performed their duties within the four walls of the parochial school that rendered monitoring difficult and, in the end, unconstitutional. It seems inherent in the pedagogical function that there will be disagreements about what is or is not “religious” and which will require an intolerable degree of government intrusion and censorship.
“What would appear to some to be essential to good citi- ' zenship might well for others border on or constitute instruction in religion. . . .
“. . . Unlike a book, a teacher cannot be inspected once so as to determine the extent and intent of his or her personal beliefs and subjective acceptance of the limitations imposed by the First Amendment.” Id., at 619.
Accord, Aguilar v. Felton, 473 U. S., at 413. See also New York v. Cathedral Academy, 434 U. S. 125, 133 (1977) (noting that the State “would have to undertake a search *651for religious meaning in every classroom examination .... The prospect of church and state litigating in court about what does or does not have religious meaning touches the very core of the constitutional guarantee against religious establishment”).
In Roemer, Tilton, and Hunt, the Court relied on “the ability of the State to identify and subsidize separate secular functions carried out at the school, without on-the-site inspections being necessary to prevent diversion of the funds to sectarian purposes,” Roemer v. Maryland Public Works Board, 426 U. S., at 765 (emphasis added), and on the fact that onetime grants require “no continuing financial relationships or dependencies, no annual audits, and no government analysis of an institution’s expenditures on secular as distinguished from religious activities.” Tilton v. Richardson, 403 U. S., at 688. AFLA grants, of course, are not simply one-time construction grants. As the majority readily acknowledges, the Secretary will have to “review the programs set up and run by the AFLA grantees[, including] a review of, for example, the educational materials that a grantee proposes to use.” Ante, at 616-617. And, as the majority intimates, monitoring the use of AFLA funds will undoubtedly require more than the “minimal” inspection “necessary to ascertain that the facilities are devoted to secular education,” Tilton, 403 U. S., at 687. Since teachers and counselors, unlike buildings, “are not necessarily religiously neutral, greater governmental surveillance would be required to guarantee that state salary aid would not in fact subsidize religious instruction.” Id., at 687-688.
V
The AFLA, without a doubt, endorses religion. Because of its expressed solicitude for the participation of religious organizations in all AFLA programs in one form or another, the statute creates a symbolic and real partnership between the clergy and the fisc in addressing a problem with substan*652tial religious overtones. Given the delicate subject matter and the impressionable audience, the risk that the AFLA will convey a message of Government endorsement of religion is overwhelming. The statutory language and the extensive record established in the District Court make clear that the problem lies in the statute and its systematically unconstitutional operation, and not merely in isolated instances of misapplication. I therefore would find the statute unconstitutional without remanding to the District Court. I trust, however, that after all its labors thus far, the District Court will not grow weary prematurely and read into the Court’s decision a suggestion that the AFLA has been constitutionally implemented by the Government, for the majority deliberately eschews any review of the facts.16 After such further *653proceedings as are now to be deemed appropriate, and after the District Court enters findings of fact on the basis of the testimony and documents entered into evidence, it may well decide, as I would today, that the AFLA as a whole indeed has been unconstitutionally applied.17
A related point on which I do agree with the majority is worth acknowledging explicitly. In his appeal to this Court, the Secretary of Health and Human Services vigorously criticized the District Court’s analysis of the AFLA on its face, asserting that it “cannot.be squared with-this Court’s explanation in United States v. Salerno, [481 U. S. 739, 745 (1987),] that in mounting a facial challenge to a legislative Act, ‘the challenger must establish that no set of circumstances exists under which the Act would.be valid.’” Brief for Federal Appellant 30. The Court, however, rejects the application of such rigid analysis in Establishment Clause cases, explaining: “As in previous cases involving facial challenges on Establishment Clause grounds, ... we assess the constitutionality of an enactment by reference to the three factors first articulated in Lemon v. Kurtzman, 403 U. S. 602 (1971).” Ante, at 602. Indeed, the Secretary’s proposed test is wholly incongruous with analysis of an Establishment Clause challenge under Lemon, which requires our examination of the purpose of the legislative enactment, as well as its primary effect' or potential for fostering excessive entanglement. Although I may differ with the majority in the application of the Lemon- analysis to the AFLA, I join it in rejecting the Secretary’s approach which would render review under the Establishment Clause a nullity. Even in a statute like the AFLA, with its solicitude for, and specific averment to, the participation of religious organizations, one eould hypothesize some “set of circumstances . . . under which the Act would be valid,” as, for example, might be the case if no religious organization ever actually applied for or participated under an AFLA grant. The Establishment Clause cannot be eviscerated by such artifice.
Of course, the manner in which the challenge is characterized does not limit the relief available. Where justified by the nature of the controversy and the evidence in the record, a federal district court may invoke broad equitable powers to pi-event continued unconstitutional activity. See Hutto v. Finney, 437 U. S. 678, 687, and n. 9 (1978); Swann v. Charlotte-Mecklenburg Board of Education, 402 U. S. 1, 15 (1971) (“[B]readth and flexibility are inherent in equitable remedies”). In Milliken v. Bradley, 433 U. S. 267 (1977), the Court reiterated that in exercising its broad equitable powers, a district court should focus on the “nature and scope of the constitutional violation,” and ensure that decrees be “remedial in nature.” Id., at 280 (emphasis omitted). On remand, therefore, as instructed by the majority, the District Court must undertake the delicate task of fashioning relief appropriate to the scope of any particular violation it discovers.
The majority finds support for its “observation^]” in the statistics for the AFLA program in fiscal 1986. See date, at 610, n. 12. Because there are some organizations that were funded in 1982, but not in 1986, and vice versa, I find' the cumulative funding figures for FY 1982-1986 more helpful. Looking at those figures, and the same group of recipients identified by the majority, I find that of approximately $53.5 million in AFLA funding, over $10 million went to the 13 organizations specifically cited in the District Court’s opinion for constitutional violations. App. 748-756. The District Court, of course, did not “engage in an exhaustive recitation of the record,’’ but made references only to “representative portions.” 657 F. Supp. 1547, 1564 (DC 1987). Another 13 organizations characterized as “religiously affiliated” in a tabulation prepared by the Department of Health and Human Services in connection with this litigation, received an additional $6 million during this period. Looking at the figures from a different perspective, a third of the approximately 100,000 “clients served” by all AFLA grantees during the 1985-1986 period received their services from the “cited” grantees, and nearly 11,000 more from the other “religiously affiliated” institutions. App. 748-756. At a minimum, these figures already demonstrate substantial constitutionally suspect funding through the AFLA, rendering the majority’s expectations unrealistic and *630unwarranted. And, because of the Government’s failure to require grantees to report on subgrant and subcontract arrangements, id., at 745, we only can speculate as to what additional public funds subsidized the religious missions of groups that the secular grantees brought in to fulfill their statutory obligation to involve religious organizations in the provision of services. See § 300z-5(a)(21)(B).
The Court leaves for the District Court on remand the “consideration of the evidence presented by appellees insofar as it sheds light on the manner in which the statute is presently being administered,” ante, at 621, conceding, as it must, that the factual record could paint a troubling picture about the true efféct of the AFLA as a whole. See Witters v. Washington Dept. of Services for the Blind, 474 U. S. 481, 488 (1986) (finding significant that “nothing in the record indicates that, if petitioner succeeds, any significant portion of the aid expended under the . . . program as a whole will end up flowing to religious education”); Aguilar v. Felton, 473 U. S. 402, 412, n. 8 (1985) (“ ‘If any significant number of the . . . schools create the risks described in Meek, Meek applies’”), quoting Felton v. Secretary, United States Dept. of Education, 739 F. 2d 48, 70 (CA2 1984); Widmar v. Vincent, 454 U. S. 263, 275 (1981) (noting absence of empirical evidence that religious groups would dominate university’s, open forum).
I fully agree with the majority’s determination that appellees have standing as taxpayers to challenge the operation- of the AFLA, ante, at 618-620, and note that appellees may yet prevail on remand.
In rejecting the claim that the AFLA leads to excessive government entanglement with religion, the. Court declines “to assume that the reli*632gious organizations which may receive grants are ‘pervasively sectarian’ in the same sense as the Court has held parochial schools to be.” Ante, at 616. With respect to the claim that the AFLA is unconstitutional at least as applied, if not on its face, the Court — apparently unsatisfied with findings the District Court already made to that very effect — instructs that on remand, appellees may show that “AFLA aid is flowing to grantees that can be considered ‘pervasively sectarian’ religious institutions, such as we have held parochial schools to be.” Ante, at 621.
The District Court observed that 9 of 17 “necessary services,” see § 300z-l(a)(4), expressly involved some sort of education, counseling, or an intimately related service. 657 F. Supp., at 1562.
Thus, for example, until discovery began in this lawsuit, St. Ann’s, a home for unmarried pregnant teenagers, operated by the Order of the Daughters of Charity and owned by the Archdiocese of Washington, D.C., purchased books containing Catholic doctrine on chastity, masturbation, homosexuality, and abortion, using AFLA funds, and distributed them to participants. See App. 336, 354-359, 362. Catholic Family Services of Amarillo, Tex., used a curriculum outline guide for AFLA-funded parent workshops with explicit theological references, as well as religious “reference” materials, including the film “Everyday Miracle,” described as “depicting the miracle of the process of human reproduction as a gift from God.” Record 155, Plaintiffs’ Appendix, Vol. IV, p. 119. The District Court concluded:
“The record demonstrates that some grantees have included explicitly religious materials, or a curriculum that indicates an intent to teach theological and secular views on- sexual conduct, in their HHS-approved grant proposals. . . . One such application, which was funded for one year, included a program designed, inter alia, ‘to communicate the Catholic diocese, Mormon (Church of Jesus Christ of Latter Day Saints) and Young Buddhist Association’s approaches to sex education.’” 657 F. Supp., at 1565-1566.
In addition to funding activity of a wholly different character, the AFLA differs from the statutes reviewed in those cases in its expressed solicitude for the participation of religious organizations. In Tilton v. Richardson, 403 U. S. 672, 675 (1971), the statute “authorized] federal grants and loans to ‘institutions of higher education’ for the construction of a wide variety of ‘academic facilities’in Hunt v. McNair, 413 U. S. 734, 736 (1973), South Carolina had established a state agency “the purpose of which [was] ‘to assist institutions for higher education in the construction, financing and refinancing of projects’. . . primarily through the issuance of revenue bonds”; in Roemer v. Maryland Public Works Board, 426 U. S. 736, 740 (1976) (plurality opinion), the State provided funding to “‘any [qualified] private institution of higher learning within the State of Maryland.’ ” The AFLA, in contrast, expressly requires applicants for grants to describe how they “will, as appropriate in the provision of services . . . (B) involve religious . . . organizations.” § 300z-5(a)(21 j(B), and the legislative history conclusively shows that Congress intended religious organizations to participate as grantees and as participants under grants awarded to other organizations. See S. Rep. No. 97-161, pp. 15-16 (1981).
The District Court’s conclusion, which I find compelling, is that the AFLA requires teaching and counseling “on matters inseparable from religious dogma.” 657 F. Supp., at 1565. This conclusion is borne out by statements of AFLA administrators and participants. For example, the Lyon County, Kan., Health Department’s grant proposal acknowledges that “[s]uch sensitive and intimate material cannot be presented without touching on . . . religious beliefs.” Record 155, Plaintiffs' Appendix, Vol. IV, p. 221. Patrick J. Sheeran, the Director of the Division of Program Development and Monitoring in the Office of Adolescent Pregnancy Programs explained;
“Broadly speaking, I find it hard to find any kind of educational or value type of program that doesn’t have some kind of basic religious or ethical foundation, and while a sex education class may be completely separate from a religious class, it might relate back to it in terms of principles that are embedded philosophically or theologically or religiously in another discipline.” App. 122.
Mr. Sheeran’s views were echoed by Dr. Paul Simmons, a Baptist clergyman and professor of Christian Ethics:
“The very purpose of religion is to transmit certain values, and those values associated with sex, marriage, chastity and abortion involve religious values and theological or doctrinal issues. In encouraging premarital chastity, it would be extremely difficult for a religiously affiliated group not to impart its own religious values and doctrinal perspectives when teaching a subject that has always been central to its religious teachings.” Id., at 597.
In any event, regardless of the efforts AFLA teachers and counselors may have undertaken in attempting to separate their religious convictions from the advice they actually dispensed to participating teenagers, the Dis*640trict Court found that “the overwhelming number of comments shows that program participants believed that these federally funded programs were also sponsored by the religious denomination.” 657 F. Supp., at 1566.
Religion plays an important role to many in our society. By enlisting its aid in combating certain social ills, while imposing the restrictions required by the First Amendment on the use of public funds to promote religion, we risk secularizing and demeaning the sacred enterprise. Whereas there is undoubtedly a role for churches of all denominations in helping prevent the problems often associated with early sexual activity and unplanned pregnancies, any attempt to confine that role within the strictures of a government-sponsored secular program can only taint the religious mission with a “corrosive secularism.” Grand Rapids School District v. Ball, 473 U. S. 373, 385 (1985). The First Amendment protects not only the State from being captured by the Church, but also protects the Church from being corrupted by the State and adopted for its purposes. A government program that provides funds for religious organizations to carry out secular tasks inevitably risks promoting “the pernicious tendency of a state subsidy to tempt religious schools to compromise their religious mission without wholly abandoning it.” Roemer v. Maryland Public Works Board, 426 U. S., at 775 (Stevens, J., dissenting); see also Lynch v. Donnelly, 465 U. S. 668, 726-727 (1984) (Blackmun, J., dissenting).
In arguing that providing “social welfare services” is categorically different from educating schoolchildren for Establishment Clause purposes, appellants relied heavily on Bradfield v. Roberts, 175 U. S. 291 (1899), a case in which the Court upheld the appropriation of money for the construction of two buildings to be part of a religiously affiliated hospital. Unlike the AFLA, however, which seeks “to promote self discipline and other prudent approaches to the problem of adolescent premarital sexual relations,” § 300z(b)(l), the Act of Congress by which the hospital at issue in Bradfield had been incorporated expressed that “ ‘the specific and limited object of its creation’ is the opening and keeping a hospital in the city of Washington for the care of such sick and invalid persons as may place themselves under the treatment and care of the corporation.” 175 U. S., at 299-300.
Employees of some grantees must follow the directives set forth in a booklet entitled “The Ethical and Religious Directives for Catholic Health Facilities,” approved by the Committee on Doctrine of the National Conference of Catholic Bishops. App. 526, 540-544. Solely because of religious dictates, some AFLA grantees teach and refer teenagers for only “natural family planning," which “has never been used successfully with teenagers,” id., at 535, and may not refer couples to programs that offer artificial methods of birth control, because those programs conflict with the teachings of the Roman Catholic Church. Id., at 407, 628. One nurse midwife working at an AFLA program was even reprimanded for contravening the hospital’s religious views on sex vdien she answered “yes” to a teenager who asked, as a medical matter, whether she could have sex during pregnancy. Id., at 552.
This vacuum is particularly noticeable when we consider the pains to which Congress went to specify other restrictions on the use of AFLA funds. For example, the AFLA expressly provides:
“Grants or payments may be made only to programs or projects which do not provide abortions or abortion counseling or referral, or which do not subcontract with or make any payment to any person who provides abortions or abortion counseling or referral, except that any such program or project may provide referral for abortion counseling to a pregnant adolescent if such adolescent and the parents or guardians of such adolescent request such referral; and grants may be made only to projects or programs which do not advocate, promote, or encourage abortion.” §300z-10.
The AFLA also sets certain conditions on funding for family planning, services, § 300z — 3(b)(1), and requires of applicants some 18 separate “assur-*644anees” covering everything from confidentiality of patient records, § 300z-5(a)(ll), to a commitment that the applicant will “make every reasonable effort... to secure from eligible persons payment for services in accordance with [structured fee] schedules,” § 300z-5(a)(16)(B). Yet nowhere in the statute is there a single restriction on the use of federal funds to promote or advance religion. See ante, at 614-615.
Appellees have challenged that presumption here, calling into question the manner in which grantees were selected and supervised. Mr. Shee-ran, the Director of the Division of Program Development and Monitoring in the Office of Adolescent Pregnancy Programs, testified that he was surprised at the lack of experience, yet high proportion of religious affiliation, among those selected to read and evaluate grant applications. App. 98. Some of the reader’s comments strongly suggest they considered religious indoctrination indispensable to achieve the AFLA’s stated purpose, see, e. g., id., at 509; Record 155, Plaintiffs’ Appendix Vol. I, pp. 354-355, and that evidence of no involvement by religious*organizations was a factor in rejecting applications, see, e. g., Record 155, Plaintiffs’ Appendix, Vol. I-A, pp. 505D, 505E, 505G; Record 155, Plaintiffs’ Appendix, Vol. I, pp. 340, 346.
Despite the clear religious mission of many applicants, pre-award investigations or admonitions against the use of AFLA funds to promote religion were minimal. Mr. Sheeran was instructed to call Catholic grantees already selected for funding, and obtain assurances that the grant money would not be used for “teaching of morals, dogmas, [or] religious principles.” App. 107. The calls lasted two or three minutes, and involved no detailed discussion of the use of church and parochial school facilities, or religious literature. Id., at 112-113.
The District Court found that the problems that should have been noted at the application stage remained uncured in implementation:
“Nor do the facts suggest that the programs in operation cured the First Amendment problems evident from these approved grant applications. At least one grantee actually included ‘spiritual counseling’ in its AFLA program. Other AFLA programs used curricula with explicitly religious *645materials. In addition, a very large number of AFLA programs took place on sites adorned with religious symbols ....
“Similarly, the record reveals that some grantees attempted to evade restrictions they perceived on AFLA-funded religious teaching by establishing programs in which an AFLA-funded staffer’s presentations would be immediately followed, in the same room and in the staffer’s presence, by a program presented by a member of a religious order and dedicated co presentation of religious views on the subject covered by the AFLA staffer” (citations omitted). 657 F. Supp., at 1566.
Indeed, the AFLA stands out among similar grant programs, precisely because of the absence of such restrictions. Cf., e. g., 20 U. S. C. § 27 (support for vocational education); 20 U. S. C. § 241-l(a)(4) (federal disaster relief for local education agencies); 20 U. S. C. § 1021(c) (assistance to college and research libraries); 20 U. S. C. § 1070e(c)(l)(B) (1982 ed., Supp. IV) (assistance to institutions of higher education); 20 U. S. C. § 1134e(g) (1982 ed., Supp. IV) (fellowships for graduate and professional study); 20 U. S. C. § 1210 (1982 ed. and Supp. IV) (grants to adult education programs); 42 U. S. C. § 2753(b)(1)(C) (college work-study grants); 42 U. S. C. § 5001(a)(2) (grants to retired senior-citizen volunteer service programs).
Justice Kennedy, joined by Justice Scalia, would further constrain the District Court’s consideration of the evidence as to how grantees spent their money, regardless of whether the grantee could be labeled “pervasively sectarian,” see ante, at 624-625, asserting that “[t]he question in an as-applied challenge is not whether the entity is of a religious character.” This statement comes without citation to authority and is contrary to the clear import of our cases. As ill-defined as the concept behind the “pervasively sectarian” label may be, this Court consistently has held, and reaffirms today, that “ ‘[a]id normally may be thought to have a primary effect of advancing religion when it flows to an institution in which religion is so pervasive that a substantial portion of its functions are subsumed in the religious mission.’ ” Ante, at 610, quoting Hunt v. McNair, 413 U. S., at 743.
See also Roemer v. Maryland Public Works Board, 426 U. S., at 758 (“[T]he question [is] whether an institution is so ‘pervasively sectarian’ that it may receive no direct state aid of any kind”). Indeed, to suggest that because a challenge is labeled “as-applied,” the character of the institution receiving the aid loses its relevance is to misunderstand the very nature of the concept of a “pervasively sectarian” institution, which is based in part on the conclusion that the secular and sectarian activities of an institution are “inextricably intertwined,” see ante, at 620, n. 16. Not surprisingly, the Court flatly rejects Justice Kennedy’s suggestion, observing that “it will be open to appellees on remand to show that AFLA aid is flowing to grantees that can be considered ‘pervasively sectarian’ religious institutions.” Ante, at 621. '
Appellees argued in the District Court, and here as cross-appellants, that the portions of the statute inviting the participation of religious organizations were not severable, and thus that the entire statute must be held unconstitutional. I take no position on this issue.