*289 Justice Marshall

delivered the opinion of the Court.

In 1980, Tennessee proposed reducing the number of annual days of inpatient hospital care covered by its state Medicaid program. The question presented is whether the effect upon the handicapped that this reduction will have is cognizable under § 504 of the Rehabilitation Act of 1973 or its implementing regulations. We hold that it is not.

I

Faced in 1980-1981 with projected state Medicaid 1 costs of $42 million more than the State’s Medicaid budget of $388 million, the directors of the Tennessee Medicaid program decided to institute a variety of cost-saving measures. Among these changes was a reduction from 20 to 14 in the number of inpatient hospital days per fiscal year that Tennessee Medicaid would pay hospitals on behalf of a Medicaid recipient. Before the new measures took effect, respondents, Tennessee Medicaid recipients, brought a class action for declaratory and injunctive relief in which they alleged, inter alia, that the proposed 14-day limitation on inpatient coverage would have a discriminatory effect on the handicapped. 2 Statistical evidence, which petitioners do not *290 dispute, indicated that in the 1979-1980 fiscal year, 27.4% of all handicapped users of hospital services who received Medicaid required more than 14 days of care, while only 7.8% of nonhandicapped users required more than 14 days of inpatient care.

Based on this evidence, respondents asserted that the reduction would violate §504 of the Rehabilitation Act of 1973, 87 Stat. 394, as amended, 29 U. S. C. §794, and its implementing regulations. Section 504 provides:

“No otherwise qualified handicapped individual. . . shall, solely by reason of his handicap, be excluded from the participation in, be denied the benefits of, or be subjected to discrimination under any program or activity receiving Federal financial assistance . . . .” 29 U. S. C. §794.

Respondents’ position was twofold. First, they argued that the change from 20 to 14 days of coverage would have a disproportionate effect on the handicapped and hence was discriminatory. 3 The second, and major, thrust of respondents’ attack was directed at the use of any annual limitation on the number of inpatient days covered, for respondents acknowledged that, given the special needs of the handicapped for medical care, any such limitation was likely to disadvantage the handicapped disproportionately. Respondents noted, however, that federal law does not require States to impose any annual durational limitation on inpatient cover *291 age, and that the Medicaid programs of only 10 States impose such restrictions. 4 Respondents therefore suggested that Tennessee follow these other States and do away with any limitation on the number of annual inpatient days covered. Instead, argued respondents, the State could limit the number of days of hospital coverage on a per-stay basis, with the number of covered days to vary depending on the recipient’s illness (for example, fixing the number of days covered for an appendectomy); the period to be covered for each illness could then be set at a level that would keep Tennessee’s Medicaid program as a whole within its budget. 5 The State’s refusal to adopt this plan was said to result in the imposition of gratuitous costs on the handicapped and thus to constitute discrimination under §504.

A divided panel of the Court of Appeals for the Sixth Circuit held that respondents had indeed established a prima facie case of a § 504 violation. Jennings v. Alexander, 715 F. 2d 1036 (1983). The majority apparently concluded that any action by a federal grantee that disparately affects the handicapped states a cause of action under §504 and its implementing regulations. Because both the 14-day rule and any annual limitation on inpatient coverage disparately *292 affected the handicapped, the panel found that a prima facie case had been made out, and the case was remanded 6 to give Tennessee an opportunity for rebuttal. According to the panel majority, the State on remand could either demonstrate the unavailability of alternative plans that would achieve the State’s legitimate cost-saving goals with a less disproportionate impact on the handicapped, or the State could offer “a substantial justification for the adoption of the plan with the greater discriminatory impact.” Id., at 1045. We granted certiorari to consider whether the type of impact at issue in this case is cognizable under §504 or its implementing regulations, 465 U. S. 1021 (1984), and we now reverse.

II

The first question the parties urge on the Court is whether proof of discriminatory animus is always required to establish a violation of §504 and its implementing regulations, or whether federal law also reaches action by a recipient of federal funding that discriminates against the handicapped by effect rather than by design. The State of Tennessee argues that §504 reaches only purposeful discrimination against the handicapped. As support for this position, the State relies heavily on our recent decision in Guardians Assn. v. Civil Service Comm’n of New York City, 463 U. S. 582 (1983).

In Guardians, we confronted the question whether Title VI of the Civil Rights Act of 1964, 42 U. S. C. § 2000d et seq., which prohibits discrimination against racial and ethnic minorities in programs receiving federal aid, reaches both *293 intentional and disparate-impact discrimination. 7 No opinion commanded a majority in Guardians, and Members of the Court offered widely varying interpretations of Title VI. Nonetheless, a two-pronged holding on the nature of the discrimination proscribed by Title VI emerged in that case. First, the Court held that Title VI itself directly reached only instances of intentional discrimination. 8 Second, the Court held that actions having an unjustifiable disparate impact on minorities could be redressed through agency regulations designed to implement the purposes of Title VI. 9 In essence, then, we held that Title VI had delegated to the agencies in the first instance the complex determination of what sorts of disparate impacts upon minorities constituted sufficiently significant social problems, and were readily *294 enough remediable, to warrant altering the practices of the federal grantees that had produced those impacts.

Guardians, therefore, does not support petitioners’ blanket proposition that federal law proscribes only intentional discrimination against the handicapped. Indeed, to the extent our holding in Guardians is relevant to the interpretation of § 504, Guardians suggests that the regulations implementing §504, upon which respondents in part rely, could make actionable the disparate impact challenged in this case. 10 Moreover, there are reasons to pause before too quickly extending even the first prong of Guardians to § 504. Cf. Consolidated Rail Corporation v. Darrone, 465 U. S. 624, 632-633, n. 13 (1984) (recognizing distinctions between Title Viand §504). 11

*295 Discrimination against the handicapped was perceived by Congress to be most often the product, not of invidious animus, but rather of thoughtlessness and indifference — of benign neglect. 12 Thus, Representative Vanik, introducing the predecessor to § 504 in the House, 13 described the treat *296 ment of the handicapped as one of the country’s “shameful oversights,” which caused the handicapped to live among society “shunted aside, hidden, and ignored.” 117 Cong. Rec. 45974 (1971). Similarly, Senator Humphrey, who introduced a companion measure in the Senate, asserted that “we can no longer tolerate the invisibility of the handicapped in America . . . .” 118 Cong. Rec. 525-526 (1972). And Senator Cranston, the Acting Chairman of the Subcommittee that drafted § 504, 14 described the Act as a response to “previous societal neglect.” 119 Cong. Rec. 5880, 5883 (1973). See also 118 Cong. Rec. 526 (1972) (statement of cosponsor Sen. Percy) (describing the legislation leading to the 1973 Act as a national commitment to eliminate the “glaring neglect” of the handicapped). 15 Federal agencies and commentators on the plight of the handicapped similarly have found that discrimination against the handicapped is primarily the result of apathetic attitudes rather than affirmative animus. 16

In addition, much of the conduct that Congress sought to alter in passing the Rehabilitation Act would be difficult if *297 not impossible to reach were the Act construed to proscribe only conduct fueled by a discriminatory intent. For example, elimination of architectural barriers was one of the central aims of the Act, see, e. g., S. Rep. No. 93-318, p. 4 (1973), yet such barriers were clearly not erected with the aim or intent of excluding the handicapped. Similarly, Senator Williams, the chairman of the Labor and Public Welfare Committee that reported out § 504, asserted that the handicapped were the victims of “[discrimination in access to public transportation” and “[discrimination because they do not have the simplest forms of special educational and rehabilitation services they need . . . 118 Cong. Rec. 3320 (1972). And Senator Humphrey, again in introducing the proposal that later became §504, listed, among the instances of discrimination that the section would prohibit, the use of “transportation and architectual barriers,” the “discriminatory effect of job qualification . . . procedures,” and the denial of “special educational assistance” for handicapped children. Id., at 525-526. These statements would ring hollow if the resulting legislation could not rectify the harms resulting from action that discriminated by effect as well as by design. 17

*298 At the same time, the position urged by respondents — that we interpret § 504 to reach all action disparately affecting the handicapped — is also troubling. Because the handicapped typically are not similarly situated to the nonhandicapped, respondents’ position would in essence require each recipient of federal funds first to evaluate the effect on the handicapped of every proposed action that might touch the interests of the handicapped, and then to consider alternatives for achieving the same objectives with less severe disadvantage to the handicapped. The formalization and policing of this process could lead to a wholly unwieldy administrative and adjudicative burden. See Note, Employment Discrimination Against the Handicapped and Section 504 of the Rehabilitation Act: An Essay on Legal Evasiveness, 97 Harv. L. Rev. 997, 1008 (1984) (describing problems with pure disparate-impact model in context of employment discrimination against the handicapped). Had Congress intended §504 to be a National Environmental Policy Act 18 for the handicapped, requiring the preparation of “Handicapped Impact *299 Statements” before any action was taken by a grantee that affected the handicapped, we would expect some indication of that purpose in the statute or its legislative history. Yet there is nothing to suggest that such was Congress’ purpose. Thus, just as there is reason to question whether Congress intended § 504 to reach only intentional discrimination, there is similarly reason to question whether Congress intended § 504 to embrace all claims of disparate-impact discrimination.

Any interpretation of § 504 must therefore be responsive to two powerful but countervailing considerations — the need to give effect to the statutory objectives and the desire to keep § 504 within manageable bounds. Given the legitimacy of both of these goals and the tension between them, we decline the parties’ invitation to decide today that one of these goals so overshadows the other as to eclipse it. While we reject the boundless notion that all disparate-impact showings constitute prima facie cases under § 504, we assume without deciding that § 504 reaches at least some conduct that has an unjustifiable disparate impact upon the handicapped. On that assumption, we must then determine whether the disparate effect of which respondents complain is the sort of disparate impact that federal law might recognize.

Ill

To determine which disparate impacts §504 might make actionable, the proper starting point is Southeastern Community College v. Davis, 442 U. S. 397 (1979), our major previous attempt to define the scope of § 504. 19 Davis involved a plaintiff with a major hearing disability who sought admission *300 to a college to be trained as a registered nurse, but who would not be capable of safely performing as a registered nurse even with full-time personal supervision. We stated that, under some circumstances, a “refusal to modify an existing program might become unreasonable and discriminatory. Identification of those instances where a refusal to accommodate the needs of a disabled person amounts to discrimination against the handicapped [is] an important responsibility of HEW.” Id.,, at 413. We held that the college was not required to admit Davis because it appeared unlikely that she could benefit from any modifications that the relevant HEW regulations required, id., at 409, and because the further modifications Davis sought — full-time, personal supervision whenever she attended patients and elimination of all clinical courses — would have compromised the essential nature of the college’s nursing program, id., at 413-414. Such a “fundamental alteration in the nature of a program” was far more than the reasonable modifications the statute or regulations required. Id., at 410. Davis thus struck a balance between the statutory rights of the handicapped to be integrated into society and the legitimate interests of federal grantees in preserving the integrity of their programs: while a grantee need not be required to make “fundamental” or “substantial” modifications to accommodate the handicapped, it may be required to make “reasonable” ones. Compare ibid, with id., at 412-413. 20

*301 The balance struck in Davis requires that an otherwise qualified handicapped individual must be provided with meaningful access to the benefit that the grantee offers. The benefit itself, of course, cannot be defined in' a way that effectively denies otherwise qualified handicapped individuals the meaningful access to which they are entitled; to assure meaningful access, reasonable accommodations in the grantee’s program or benefit may have to be made. 21 In this *302 case, respondents argue that the 14-day rule, or any annual durational limitation, denies meaningful access to Medicaid services in Tennessee. We examine each of these arguments in turn.

A

The 14-day limitation will not deny respondents meaningful access to Tennessee Medicaid services or exclude them from those services. The new limitation does not invoke criteria that have a particular exclusionary effect on the handicapped; the reduction, neutral on its face, does not distinguish between those whose coverage will be reduced and those whose coverage will not on the basis of any test, judgment, or trait that the handicapped as a class are less capable of meeting or less likely of having. Moreover, it cannot be argued that “meaningful access” to state Medicaid services will be denied by the 14-day limitation on inpatient coverage; nothing in the record suggests that the handicapped in Tennessee will be unable to benefit meaningfully from the coverage they will receive under the 14-day rule. 22 The reduction in inpatient coverage will leave both handicapped and nonhandicapped Medicaid users with identical and effective hospital services fully available for their use, with both classes of users subject to the same durational limitation. The 14-day limitation, therefore, does not exclude the handicapped from or deny them the benefits of the 14 days of care the State has chosen to provide. Cf. Jefferson v. Hackney, 406 U. S. 535 (1972).

To the extent respondents further suggest that their greater need for prolonged inpatient care means that, to provide meaningful access to Medicaid services, Tennessee must single out the handicapped for more than 14 days of *303 coverage, the suggestion is simply unsound. At base, such a suggestion must rest on the notion that the benefit provided through state Medicaid programs is the amorphous objective of “adequate health care.” But Medicaid programs do not guarantee that each recipient will receive that level of health care precisely tailored to his or her particular needs. Instead, the benefit provided through Medicaid is a particular package of health care services, such as 14 days of inpatient coverage. That package of services has the general aim of assuring that individuals will receive necessary medical care, but the benefit provided remains the individual services offered — not “adequate health care.”

The federal Medicaid Act makes this point clear. The Act gives the States substantial discretion to choose the proper mix of amount, scope, and duration limitations on coverage, as long as care and services are provided in “the best interests of the recipients.” 42 U. S. C. § 1396a(a)(19). The District Court found that the 14-day limitation would fully serve 95% of even handicapped individuals eligible for Tennessee Medicaid, and both lower courts concluded that Tennessee’s proposed Medicaid plan would meet the “best interests” standard. That unchallenged conclusion 23 indicates that Tennessee is free, as a matter of the Medicaid Act, to choose to define the benefit it will be providing as 14 days of inpatient coverage.

Section 504 does not require the State to alter this definition of the benefit being offered-simply to meet the reality that the handicapped have greater medical needs. To conclude otherwise would be to find that the Rehabilitation Act requires States to view certain illnesses, i. e., those *304 particularly affecting the handicapped, as more important than others and more worthy of cure through government subsidization. Nothing in the legislative history of the Act supports such a conclusion. Cf. Doe v. Colautti, 592 F. 2d 704 (CA3 1979) (State may limit covered-private-inpatient-pyschiatric care to 60 days even though State sets no limit on duration of coverage for physical illnesses). Section 504 seeks to assure evenhanded treatment and the opportunity for handicapped individuals to participate in and benefit from programs receiving federal assistance. Southeastern Community College v. Davis, 442 U. S. 397 (1979). The Act does not, however, guarantee the handicapped equal results from the provision of state Medicaid, even assuming some measure of equality of health could be constructed. Ibid.

Regulations promulgated by the Department of Health and Human Services (HHS) pursuant to the Act further support this conclusion. 24 These regulations state that recipients of federal funds who provide health services cannot “provide a qualified handicapped person with benefits or services that are not as effective (as defined in § 84.4(b)) as the benefits or services provided to others.” 45 CFR §84.52(a)(3)(1984). The regulations also prohibit a recipient of federal funding from adopting “criteria or methods of administration that *305 have the purpose or effect of defeating or substantially impairing accomplishment of the objectives of the recipient’s program with respect to the handicapped.” 45 CFR § 84.4 (b) (4) (ii) (1984). 25

While these regulations, read in isolation, could be taken to suggest that a state Medicaid program must make the handicapped as healthy as the nonhandicapped, other regulations reveal that HHS does not contemplate imposing such a requirement. Title 45 CFR § 84.4(b)(2)(1984), referred to in the regulations quoted above, makes clear that

“[f]or purposes of this part, aids, benefits, and services, to be equally effective, are not required to produce the identical result or level of achievement for handicapped and nonhandicapped persons, but must afford handicapped persons equal opportunity to obtain the same result, to gain the same benefit, or to reach the same level of achievement. . . .”

This regulation, while indicating that adjustments to existing programs are contemplated, 26 also makes clear that *306 Tennessee is not required to assure that its handicapped Medicaid users will be as healthy as its nonhandicapped users. Thus, to the extent respondents are seeking a distinct durational limitation for the handicapped, Tennessee is entitled to respond by asserting that the relevant benefit is 14 days of coverage. Because the handicapped have meaningful and equal access to that benefit, Tennessee is not obligated to reinstate its 20-day rule or to provide the handicapped with more than 14 days of inpatient coverage.

B

We turn next to respondents’ alternative contention, a contention directed not at the 14-day rule itself but rather at Tennessee’s Medicaid plan as a whole. Respondents argue that the inclusion of any annual durational limitation on inpatient coverage in a state Medicaid plan violates §504. The thrust of this challenge is that all annual durational limitations discriminate against the handicapped because (1) the effect of such limitations falls most heavily on the handicapped and because (2) this harm could be avoided by the choice of other Medicaid plans that would meet the State’s budgetary constraints without disproportionately disadvantaging the handicapped. Viewed in this light, Tennessee’s current plan is said to inflict a gratuitous harm on the handicapped that denies them meaningful access to Medicaid services.

Whatever the merits of this conception of meaningful access, it is clear that §504 does not require the changes respondents seek. In enacting the Rehabilitation Act and in subsequent amendments, 27 Congress did focus on several *307 substantive areas — employment, 28 education, 29 and the elimination of physical barriers to access 30 — in which it considered the societal and personal costs of refusals to provide meaningful access to the handicapped to be particularly high. 31 But nothing in the pre- or post-1973 legislative discussion of § 504 suggests that Congress desired to make major inroads on the States’ longstanding discretion to choose the proper mix of amount, scope, and duration limitations on services covered by state Medicaid, see Beal v. Doe, 432 U. S. 438, 444 (1977). And, more generally, we have already stated, supra, at 298-299, that §504 does not impose a general NEPA-like requirement on federal grantees. 32

*308 The costs of such a requirement would be far from minimal, and thus Tennessee's refusal to pursue this course does not, as respondents suggest, inflict a “gratuitous” harm on the handicapped. On the contrary, to require that the sort of broad-based distributive decision at issue in this case always be made in the way most favorable, or least disadvantageous, to the handicapped, even when the same benefit is meaningfully and equally offered to them, would be to impose a virtually unworkable requirement on state Medicaid administrators. Before taking any across-the-board action affecting Medicaid recipients, an analysis of the effect of the proposed change on the handicapped would have to be prepared. Presumably, that analysis would have to be further broken down by class of handicap — the change at issue here, for example, might be significantly less harmful to the blind, who use inpatient services only minimally, than to other subclasses of handicapped Medicaid recipients; the State would then have to balance the harms and benefits to various groups to determine, on balance, the extent to which the action disparately impacts the handicapped. In addition, respondents offer no reason that similar treatment would not have to be accorded other groups protected by statute or regulation from disparate-impact discrimination.

It should be obvious that administrative costs of implementing such a regime would be well beyond the accommodations that are required under Davis. As a result, Tennessee need not redefine its Medicaid program to eliminate *309 durational limitations on inpatient coverage, even if in doing so the State could achieve its immediate fiscal objectives in a way less harmful to the handicapped.

IV

The 14-day rule challenged in this case is neutral on its face, is not alleged to rest on a discriminatory motive, and does not deny the handicapped access to or exclude them from the particular package of Medicaid services Tennessee has chosen to provide. The State has made the same benefit — 14 days of coverage — equally accessible to both handicapped and nonhandicapped persons, and the State is not required to assure the handicapped “adequate health care” by providing them with more coverage than the nonhandicapped. In addition, the State is not obligated to modify its? Medicaid program by abandoning reliance on annual durational limitations on inpatient coverage. Assuming, then, that §504 or its implementing regulations reach some claims of disparate-impact discrimination, the effect of Tennessee’s reduction in annual inpatient coverage is not among them. For that reason, the Court of Appeals erred in holding that respondents had established a prima facie violation of § 504. The judgment below is accordingly reversed.

It is so ordered.

Justice Stevens

announced the judgment of the Court and delivered an opinion, in which Justice Marshall, Justice Blackmun, and Justice Powell join.

This case presents the question whether certain regulations governing the provision of health care to handicapped infants are authorized by § 504 of the Rehabilitation Act of 1973. That section provides, in part:

*613“No otherwise qualified handicapped individual . . . shall, solely by reason of his handicap, be excluded from the participation in, be denied the benefits of, or be subjected to discrimination under any program or activity receiving Federal financial assistance.” 87 Stat. 394, 29 U. S. C. §794.1

I

The American Medical Association, the American Hospital Association, and several other respondents2 challenge the validity of Final Rules promulgated on January 12, 1984, by the Secretary of the Department of Health and Human Services.3 These Rules establish “Procedures relating to health care for handicapped infants,” and in particular require the posting of informational notices, authorize expedited access to records and expedited compliance actions, and command state child protective services agencies to “prevent instances of unlawful medical neglect of handicapped infants.” 45 CFR §84.55 (1985).

Although the Final Rules comprise six parts, only the four mandatory components are challenged here.4 Subsection (b) *614is entitled “Posting of informational notice” and requires every “recipient health care provider that provides health care services to infants in programs or activities receiving *615Federal financial assistance” — a group to which we refer generically as “hospitals” — to post an informational notice in one of two approved forms. 45 CFR § 84.55(b) (1985). Both forms include a statement that § 504 prohibits discrimination on the basis of handicap, and indicate that because of this prohibition “nourishment and medically beneficial treatment (as determined with respect for reasonable medical judgments) should not be withheld from handicapped infants solely on the basis of their present or anticipated mental or physical impairments.” 45 CFR §§ 84.55(b)(3), (4) (1985). The notice’s statement of the legal requirement does not distinguish between medical care for which parental consent has been obtained and that for which it has not. The notice must identify the telephone number of the appropriate child protective services agency and, in addition, a toll-free number for the Department that is available 24 hours a day. Ibid. Finally, the notice must state that the “identity of callers will be kept confidential” and that federal law prohibits retaliation “against any person who provides information about possible violations.” Ibid.

Subsection (c), which contains the second mandatory requirement, sets forth “Responsibilities of recipient state child protective services agencies.” Subsection (c) does not mention § 504 (or any other federal statute) and does not even use the word “discriminate.” It requires every designated agency to establish and maintain procedures to ensure that *616“the agency utilizes its full authority pursuant to state law to prevent instances of unlawful medical neglect of handicapped infants.” 45 CFR § 84.55(c)(1). Mandated procedures must include (1) “[a] requirement thát health care providers report on a timely basis . . . known or suspected instances of unlawful medical neglect of handicapped infants,” §84.55(c)(l)(i); (2) a method by which the state agency can receive timely reports of such cases, § 84.55(c)(1)(h); (3) “immediate” review of those reports, including “on-site investigation,” where appropriate, §84.55(c)(l)(iii); (4) protection of “medically neglected handicapped infants” including, where appropriate, legal action to secure “timely court order[s] to compel the provision of necessary nourishment and medical treatment,” §84.55(c)(l)(iv); and (5) “[tjimely notification” to HHS of every report of “suspected unlawful medical neglect” of handicapped infants. The preamble to the Final Rules makes clear that this subsection applies “where a refusal to provide medically beneficial treatment is a result, not of decisions by a health care provider, but of decisions by parents.” 49 Fed. Reg. 1627 (1984).

The two remaining mandatory regulations authorize “[expedited access to records” and “[expedited action to effect compliance.” 45 CFR §§ 84.55(d), (e) (1985). Subsection (d) provides broadly for immediate access to patient records on a 24-hour basis, with or without parental consent, “when, in the judgment of the responsible Department official, immediate access is necessary to protect the life or health of a handicapped individual.” § 84.55(d). Subsection (e) likewise dispenses with otherwise applicable requirements of notice to the hospital “when, in the judgment of the responsible Department official, immediate action to effect compliance is necessary to protect the life or health of a handicapped individual.” § 84.55(e). The expedited compliance provision is intended to allow “the government [to] see[k] a temporary restraining order to sustain the life of a handicapped infant in *617imminent danger of death.” 49 Fed. Reg. 1628 (1984). Like the provision affording expedited access to records, it applies without regard to whether parental consent to treatment has been withheld or whether the matter has already been referred to a state child protective services agency.

II

The Final Rules represent the Secretary’s ultimate response to an April 9, 1982, incident in which the parents of a Bloomington, Indiana, infant with Down’s syndrome and other handicaps refused consent to surgery to remove an esophageal obstruction that prevented oral feeding. On April 10, the hospital initiated judicial proceedings to override the parents’ decision, but an Indiana trial court, after holding a hearing the same evening, denied the requested relief. On April 12 the court asked the local Child Protection Committee to review its decision. After conducting its own hearing, the Committee found no reason to disagree with the court’s ruling.5 The infant died six days after its birth.

Citing “heightened public concern” in the aftermath of the Bloomington Baby Doe incident, on May 18, 1982, the director of the Department’s Office of Civil Rights, in response to a directive from the President, “remind[ed]” health care providers receiving federal financial assistance that newborn in*618fants with handicaps such as Down’s syndrome were protected by § 504. 47 Fed. Reg. 26027 (1982).6

This notice was followed, on March 7, 1983, by an “Interim Final Rule” contemplating a “vigorous federal role.” 48 Fed. Reg. 9630. The Interim Rule required health care providers receiving federal financial assistance to post “in a conspicuous place in each delivery ward, each maternity ward, each pediatric ward, and each nursery, including each intensive care nursery” a notice advising of the applicability of § 504 and the availability of a telephone “hotline” to report suspected violations of the law to HHS. Id., at 9631. Like the Final Rules, the Interim Rule also provided for expedited compliance actions and expedited access to records and facilities when, “in the judgment of the responsible Department official,” immediate action or access was “necessary to protect the life or health of a handicapped individual.” Id., at 9632. The Interim Rule took effect on March 22.

On April 6, 1983, respondents American Hospital Association et al. filed a complaint in the Federal District Court for the Southern District of New York seeking a declaration that the Interim Final Rule was invalid and an injunction against its enforcement. Little more than a week later, on April 14, in a similar challenge brought by the American Academy of Pediatrics and other medical institutions, the Federal District Court for the District of Columbia declared the Interim Final Rule “arbitrary and capricious and promulgated in violation of the Administrative Procedure Act.” American Academy of Pediatrics v. Heckler, 561 F. Supp. 395, 404 (1983). The District Judge in that case “conclude[d] that haste and inexperience ha[d] resulted in agency action based on inadequate consideration” of several relevant concerns *619and, in the alternative, found that the Secretary had improperly failed to solicit public comment before issuing the Rule. Id., at 399-401.

On July 5, 1983, the Department issued new “Proposed Rules” on which it invited comment. Like the Interim Final Rule, the Proposed Rules required hospitals to post informational notices in conspicuous places and authorized expedited access to records to be followed, if necessary, by expedited compliance action. 48 Fed. Reg. 30851. In a departure from the Interim Final Rule, however, the Proposed Rules required federally assisted state child protective services agencies to utilize their “full authority pursuant to State law to prevent instances of medical neglect of handicapped infants.” Ibid. Mandated procedures mirrored those contained in the Final Rules described above. Ibid. The preamble and appendix to the Proposed Rules did not acknowledge that hospitals and physicians lack authority to perform treatment to which parents have not given their consent.7

*620After the period for notice and comment had passed, HHS, on December 30, 1983, promulgated the Final Rules and announced that they would take effect on February 13, 1984. On March 12 of that year respondents American Hospital Association et al. amended their complaint and respondents American Medical Association et al. filed suit to declare the new regulations invalid and to enjoin their enforcement. The actions were consolidated in the Federal District Court for the Southern District of New York, which awarded the requested relief on the authority of the decision of the United States Court of Appeals for the Second Circuit in United States v. University Hospital, 729 F. 2d 144 (1984). American Hospital Assn. v. Heckler, 585 F. Supp. 541 (1984); App. to Pet. for Cert. 50a. On appeal, the parties agreed that the reasoning of the Court of Appeals in University Hospital, if valid, required a judgment against the Government in this case.8 In accordance with its earlier decision, the Court of Appeals summarily affirmed the District Court. 694 F. 2d 676 (1984). Since the judgment here thus rests entirely on the reasoning of University Hospital, it is appropriate to examine that case now.

Ill

On October 11, 1983, after the Department’s Interim Final Rule had been declared invalid but before it had promulgated the Final Rules challenged here, a child with multiple congenital defects known as “Baby Jane Doe” was born in Long *621Island, New York, and was promptly transferred to University Hospital for corrective surgery. After consulting with physicians and other advisers, the parents decided to forgo corrective surgery that was likely to prolong the child’s life, but would not improve many of her handicapping conditions.

On October 16, 1983, an unrelated attorney named Wash-burn filed suit in the New York Supreme Court, seeking the appointment of a guardian ad litem for the infant who would direct the hospital to perform the corrective surgery. The trial court granted that relief on October 20, but was reversed the following day by the Appellate Division which found that the “concededly concerned and loving parents” had “chosen one course of appropriate medical treatment over another” and made an informed decision that was “in the best interest of the infant.” Weber v. Stony Brook Hospital, 95 App. Div. 2d 587, 589, 467 N. Y. S. 2d 685, 687 (per curiam). On October 28, the New York Court of Appeals affirmed, but on the ground that the trial court should not have entertained a petition to initiate child neglect proceedings by a stranger who had not requested the aid of the responsible state agency. Weber v. Stony Brook Hospital, 60 N. Y. 2d 208, 211-213, 456 N. E. 2d 1186, 1187-1188 (per curiam).

While the state proceedings were in progress, on October 19, HHS received a complaint from a “private citizen” that Baby Jane Doe was being discriminatorily denied medically indicated treatment. HHS promptly referred this complaint to the New York State Child Protective Service. (The agency investigated the charge of medical neglect and soon thereafter concluded that there was no cause for state intervention.) In the meantime, before the State Child Protective Service could act, HHS on October 22, 1983, made repeated requests of the hospital to make its records available for inspection in order to determine whether the hospital was in compliance with § 504. The hospital refused the requests *622and advised HHS that the parents had not consented to a release of the records.

Subsequently, on November 2, 1983, the Government filed suit in Federal District Court invoking its general authority to enforce §504 and 45 CFR §84.61 (1985), a regulation broadly authorizing access to information necessary to ascertain compliance. The District Court allowed the parents to intervene as defendants, expedited the proceeding, and ruled against the Government. It reasoned that the Government had no right of access to information because the record clearly established that the hospital had not violated the statute. United States v. University Hospital, State Univ. of N. Y. at Stony Brook, 575 F. Supp. 607, 614 (EDNY). Since the uncontradicted evidence established that the hospital “ha[d] at all times been willing to perform the surgical procedures in question, if only the parents . . . would consent,” the hospital “failed to perform the surgical procedures in question, not because Baby Jane Doe [wa]s handicapped, but because her parents ha[d] refused to consent.” Ibid.

The Court of Appeals affirmed. In an opinion handed down on February 23, 1984, six weeks after promulgation of the Final Rules, it agreed with the District Court that “an agency is not entitled to information sought in an investigation that ‘overreaches the authority Congress has given.’” 729 F. 2d, at 150 (quoting Oklahoma Press Publishing Co. v. Walling, 327 U. S. 186, 217 (1946)). It further held that although Baby Jane Doe was a “handicapped individual,” she was not “otherwise qualified” within the meaning of §504 because “where medical treatment is at issue, it is typically the handicap itself that gives rise to, or at least contributes to the need for services”; as a result “the ‘otherwise qualified’ criterion of section 504 cannot be meaningfully applied to a medical treatment decision.” 729 F. 2d, at 156. For the same reason, the Court of Appeals rejected the Government’s argument that Baby Jane Doe had been “subjected to discrimination” under § 504: “Where the handicapping condi*623tion is related to the condition(s) to be treated, it will rarely, if ever, be possible to say with certainty that a particular decision was ‘discriminatory’.” Id., at 157. The difficulty of applying §504 to individual medical treatment decisions confirmed the Court of Appeals in its view that “[CJongress never contemplated that section 504 of the Rehabilitation Act would apply to treatment decisions involving defective newborn infants when the statute was enacted in 1973, when it was amended in 1974, or at any subsequent time.” Id., at 161. It therefore rejected “the far-reaching position advanced by the government in this case” and concluded that until Congress had. spoken, “it would be an unwarranted exercise of judicial power to approve the type of investigation that ha[d] precipitated this lawsuit.” Ibid.

Judge Winter dissented. He pointed out that §504 was patterned after § 601 of the Civil Rights Act of 1964, which prohibits discrimination on the basis of race in federally funded programs, and asserted that a refusal to provide medical treatment because of a person’s handicapping condition is as clearly covered by § 504 as a refusal based on a person’s race is covered by § 601:

“A judgment not to perform certain surgery because a person is black is not a bona fide medical judgment. So too, a decision not to correct a life threatening digestive problem because an infant has Down’s Syndrome is not a bona fide medical judgment. The issue of parental authority is also quickly disposed of. A denial of medical treatment to an infant because the infant is black is not legitimated by parental consent.” Id., at 162.

The Government did not file a certiorari petition in University Hospital. It did, however, seek review of the judgment in this case. We granted certiorari, 472 U.. S. 1016 (1985), and we now affirm.

*624h — I C

The Solicitor General is correct that “handicapped individual” as used in § 504 includes an infant who is born with a congenital defect. If such an infant is “otherwise qualified” for benefits under a program or activity receiving federal financial assistance, § 504 protects him from discrimination “solely by reason of his handicap.”9 It follows, under our decision in Alexander v. Choate, 469 U. S. 287, 301 (1985), that handicapped infants are entitled to “meaningful access” to medical services provided by hospitals, and that a hospital rule or state policy denying or limiting such access would be subject to challenge under § 504.

However, no such rule or policy is challenged, or indeed has been identified, in this case. Nor does this case, in contrast to the University Hospital litigation, involve a claim that any specific individual treatment decision violates § 504. This suit is not an enforcement action, and as a consequence it is not necessary to determine whether § 504 ever applies to individual medical treatment decisions involving handicapped infants. Respondents brought this litigation to challenge the four mandatory components of the Final Rules on their face,10 and the Court of Appeals’ judgment which we review merely affirmed the judgment of the District Court which “declared invalid and enjoined enforcement of [the final] regulations, *625purportedly promulgated pursuant to section 504 of the Rehabilitation Act of 1973, 29 U. S. C. §794 (1982).” App. to Pet. for Cert. 2a.11 The specific question presented by this *626case, then, is whether the four mandatory provisions of the Final Rules are authorized by § 504.

V

It is an axiom of administrative law that an agency’s explanation of the basis for its decision must include “a ‘rational connnection between the facts found and the choice made.’” Motor Vehicle Mfrs. Assn. v. State Farm Mut. Automobile Ins. Co., 463 U. S. 29, 43 (1983) (quoting Burlington Truck Lines, Inc. v. United States, 371 U. S. 156, 168 (1962)).12 Agency deference has not come so far that we will uphold regulations whenever it is possible to “conceive a basis” for administrative action. To the contrary, the “presumption of *627regularity afforded an agency in fulfilling its statutory mandate” is not equivalent to “the minimum rationality a statute must bear in order to withstand analysis under the Due Process Clause.” Motor Vehicle Mfrs. Assn. v. State Farm Mut. Automobile Ins. Co., 463 U. S., at 43, n. 9. Thus, the mere fact that there is “some rational basis within the knowledge and experience of the [regulators],” United States v. Carolene Products Co., 304 U. S. 144, 152 (1938) (footnote omitted), under which they “might have concluded” that the regulation was necessary to discharge their statutorily authorized mission, Williamson v. Lee Optical Co., 348 U. S. 483, 487 (1955), will not suffice to validate agency decision-making. See Industrial Union Dept. v. American Petroleum Inst., 448 U. S. 607, 639-659 (1980) (opinion of Stevens, J.); Burlington Truck Lines, Inc. v. United States, 371 U. S. 156, 169 (1962). Our recognition of Congress’ need to vest administrative agencies with ample power to assist in the difficult task of governing a vast and complex industrial Nation carries with it the correlative responsibility of the agency to explain the rationale and factual basis for its decision, even though we show respect for the agency’s judgment in both.

Before examining the Secretary’s reasons for issuing the Final Rules, it is essential to understand the pre-existing state-law framework governing the provision of medical care to handicapped infants. In broad outline, state law vests decisional responsibility in the parents, in the first instance, subject to review in exceptional cases by the State acting as parens patriae.13 Prior to the regulatory activity culminat*628ing in the Final Rules, the Federal Government was not a participant in the process of making treatment decisions for newborn infants. We presume that this general framework was familiar to Congress when it enacted § 504. See Cannon v. University of Chicago, 441 U. S. 677, 696-697 (1979). It therefore provides an appropriate background for evaluating the Secretary’s action in this case.

The Secretary has identified two possible categories of violations of §504 as justifications for federal oversight of handicapped infant care. First, he contends that a hospital’s refusal to furnish a handicapped infant with medically beneficial treatment “solely by reason of his handicap” constitutes unlawful discrimination. Second, he maintains that a hospital’s failure to report eases of suspected medical neglect to a *629state child protective services agency may also violate the statute. We separately consider these two possible bases for the Final Rules.14

*630I — l >

In the immediate aftermath of the Bloomington Baby Doe incident, the Secretary apparently proceeded on the assumption that a hospital’s statutory duty to provide treatment to handicapped infants was unaffected by the absence of parental consent. See supra, at 617-619. He has since abandoned that view. Thus, the preamble to the Final Rules correctly states that when “a non-treatment decision, no matter how discriminatory, is made by parents, rather than by the hospital, section 504 does not mandate that the hospital unilaterally overrule the parental decision and provide treatment notwithstanding the lack of consent.” 49 Fed. Reg. 1631 (1984). A hospital’s withholding of treatment when no parental consent has been given cannot violate § 504, for without the consent of the parents or a surrogate decisionmaker the infant is neither “otherwise qualified” for treatment nor has he been denied care “solely by reason of his handicap.”15 Indeed, it would almost certainly be a tort as a matter of state law to operate on an infant without parental consent. This analysis makes clear that the Government’s heavy reliance on the analogy to race-based refusals which violate § 601 *631of the Civil Rights Act is misplaced. If, pursuant to its normal practice, a hospital refused to operate on a black child whose parents had withheld their consent to treatment, the hospital’s refusal would not be based on the race of the child even if it were assumed that the parents based their decision entirely on a mistaken assumption that the race of the child made the operation inappropriate.

Now that the Secretary has acknowledged that a hospital has no statutory treatment obligation in the absence of parental consent, it has become clear that the Final Rules are not needed to prevent hospitals from denying treatment to handicapped infants. The Solicitor General concedes that the administrative record contains no evidence that hospitals have ever refused treatment authorized either by the infant’s parents or by a court order. Tr. of Oral Arg. 8. Even the Secretary never seriously maintained that posted notices, “hotlines,” and emergency on-site investigations were necessary to process complaints against hospitals that might refuse treatment requested by parents. The parental interest in calling such a refusal to the attention of the appropriate authorities adequately vindicates the interest in enforcement of § 504 in such cases, just as that interest obviates the need for a special regulation to deal with refusals to provide treatment on the basis of race which may violate §601 of the Civil Rights Act.

The Secretary’s belated recognition of the effect of parental nonconsent is important, because the supposed need for federal monitoring of hospitals’ treatment decisions rests entirely on instances in which parents have refused their consent. Thus, in the Bloomington, Indiana, case that precipitated the Secretary’s enforcement efforts in this area,16 as *632well as in the University Hospital case that provided the basis for the summary affirmance in the case now before us,17 the hospital’s failure to perform the treatment at issue rested on the lack of parental consent. The Secretary’s own summaries of these cases establish beyond doubt that the respective hospitals did not withhold medical care on the basis of handicap and therefore did not violate § 504; as a result, they provide no support for his claim that federal regulation is needed in order to forestall comparable cases in the future.

The Secretary’s initial failure to recognize that withholding of consent by parents does not equate with discriminatory denial of treatment by hospitals likewise undermines the Secretary’s findings in the preamble to his proposed rulemaking. In that statement, the Secretary cited four sources in support of the claim that “Section 504 [is] not being uniformly followed.” 48 Fed. Reg. 30847 (1983). None of the cited examples, however, suggests that recipients of federal financial assistance, as opposed to parents, had withheld medical care on the basis of handicap.18

*633Notwithstanding the ostensible recognition in the preamble of the effect of parental nonconsent on a hospital’s obligation to provide care, in promulgating the Final Rules the Secretary persisted in relying on instances in which parents had refused consent to support his claim that, regardless of its “magnitude,” there is sufficient evidence of “illegality” to justify “establishing basic mechanisms to allow for effective enforcement of a clearly applicable statute.” 49 Fed. Reg. 1645 (1984). We have already discussed one source of this evidence — “the several specific cases cited in the preamble to the proposed rule.” Ibid. Contrary to the Secretary’s belief, these cases do not “support the proposition that handicapped infants may be subjected to unlawful discrimination.” Ibid. In addition to the evidence relied on in prior notices, the Secretary included a summary of the 49 “Infant Doe *634cases” that the Department had processed before December 1, 1983.19 Curiously, however, by the Secretary’s own admission none of the 49 cases had “resulted in a finding of discriminatory withholding of medical care.” Id., at 1649. In fact, in the entire list of 49 cases there is no finding that a hospital failed or refused to provide treatment to a handicapped infant for which parental consent had been given.20

Notwithstanding this concession, the Secretary “believes three of these cases demonstrate the utility of the procedural *635mechanisms called for in the final rules.” Ibid. Accord, ibid. (“[T]hese cases provide additional documentation of the need for governmental involvement and the appropriateness of the procedures established by the final rules”). However, these three cases, which supposedly provide the strongest support for federal intervention, fail to disclose any discrimination against handicapped newborns in violation of § 504. For example, in Robinson, Illinois, the Department conducted an on-site investigation when it learned that the “hospital (at the parents’ request) failed to perform necessary surgery.” Id., at 1646 (emphasis added). After “[t]he parents refused consent for surgery,” “the hospital referred the matter to state authorities, who accepted custody of the infant and arranged for surgery and adoption,” all “in compliance with section 504.” Ibid. The Secretary concluded that “the involvement of the state child protective services agency,” at the behest of the hospital, “was the most important element in bringing about corrective surgery for the infant. . . . Had there been no governmental involvement in the case, the outcome might have been much less favorable.” Id., at 1649 (emphasis added).21

The Secretary’s second example illustrates with even greater force the effective and nondiscriminatory functioning of state mechanisms and the consequent lack of support for federal intervention. In Daytona Beach, Florida, the Department’s hotline received a complaint of medical neglect of a handicapped infant; immediate contact with the hospital and state agency revealed that “the parents did not consent to surgery” for the infant. Id., at 1648. Notwithstanding this information, which was confirmed by both the hospital and the state agency, and despite the fact that the state agency had “obtained a court order to provide surgery” the day before HHS was notified, the Department conducted an *636on-site investigation. Ibid. In the third case, in Colorado Springs, Colorado, the Department intervened so soon after birth that “the decisionmaking process was in progress at the time the OCR [Office of Civil Rights] inquiry began,” and “it is impossible to say the surgery would not have been provided without this involvement.” Id., at 1649. “However,” the Secretary added, “the involvement of OCR and the OCR medical consultant was cooperatively received by the hospital and apparently constructive.” Ibid.

In sum, there is nothing in the administrative record to justify the Secretary’s belief that “discriminatory withholding of medical care” in violation of § 504 provides any support for federal regulation: In two of the cases (Robinson, Illinois, and Daytona Beach, Florida), the hospital’s refusal was based on the absence of parental consent, but the parents’ decision was overridden by state authorities and the operation was performed; in the third case (Colorado Springs, Colorado) it is not clear whether the parents would have given their consent or not, but the corrective surgery was in fact performed.22

*637VII

As a backstop to his manifestly incorrect perception that withholding of treatment in accordance with parental instructions necessitates federal regulation, the Secretary contends that a hospital’s failure to report parents’ refusals to consent to treatment violates §504, and that past breaches of this kind justify federal oversight.

By itself, § 504 imposes no duty to report instances of medical neglect — that undertaking derives from state-law reporting obligations or a hospital’s own voluntary practice. Although a hospital’s selective refusal to report medical neglect of handicapped infants might violate §504,23 the Secretary *638has failed to point to any specific evidence that this has occurred. The 49 actual investigations summarized in the preamble to the Final Rules do not reveal any case in which a hospital either failed, or was accused of failing, to make an appropriate report to a state agency.24 Nor can we accept the Solicitor General’s invitation to infer discriminatory nonreporting from the studies cited in the Secretary’s proposed rulemaking. Even assuming that cases in which parents have withheld consent to treatment for handicapped infants have gone unreported, that fact alone would not prove *639that the hospitals involved had discriminated on the basis of handicap rather than simply failed entirely to discharge their state-law reporting obligations, if any, a matter which lies wholly outside the nondiscrimination mandate of § 504.

The particular reporting mechanism chosen by the Secretary — indeed the entire regulatory framework imposed on state child protective services agencies — departs from the nondiscrimination mandate of §504 in a more fundamental way. The mandatory provisions of the Final Rules omit any direct requirement that hospitals make reports when parents refuse consent to recommended procedures.25 Instead, the Final Rules command state agencies to require such reports, regardless of the state agencies’ own reporting requirements (or lack thereof). 45 CFR §84.55(c)(l)(i) (1985). Far from merely preventing state agencies from remaining calculatedly indifferent to handicapped infants while they tend to the needs of the similarly situated nonhandicapped, the Final Rules command state agencies to utilize their “full authority” to “prevent instances of unlawful medical neglect of handicapped infants.” § 84.55(c)(1). The Rules effectively make medical neglect of handicapped newborns a state investigative priority, possibly forcing state agencies to shift scarce resources away from other enforcement activities — perhaps even from programs designed to protect handicapped children outside hospitals. The Rules also order state agencies to “immediately]” review reports from hospitals, §84.55(c)(l)(iii), to conduct “on-site investigation[s],” ibid., and to take legal action “to compel the provision of necessary nourishment and medical treatment,” *640§84.55(c)(l)(iv) — all without any regard to the procedures followed by state agencies in handling complaints filed on behalf of nonhandicapped infants. These operating procedures were imposed over the objection of several state child protective services agencies that the requirement that they turn over reports to HHS “conflicts with the confidentiality requirements of state child abuse and neglect statutes,” 49 Fed. Reg. 1627 (1984) — thereby requiring under the guise of nondiscrimination a service which state law denies to the nonhandicapped.26

The complaint-handling process the Secretary would impose on unwilling state agencies is totally foreign to the authority to prevent discrimination conferred on him by § 504. “Section 504 seeks to assure evenhanded treatment,” Alexander v. Choate, 469 U. S., at 304; “neither the language, purpose, nor history of § 504 reveals an intent to impose an affirmative-action obligation” on recipients of federal financial assistance, Southeastern Community College v. Davis, 442 U. S. 397, 411 (1979).27 The Solicitor General also recognizes that §504 is concerned with discrimination and with discrimination alone. In his attempt to distinguish the Secretary’s 1976 determination that it “is beyond the authority of section 504” to promulgate regulations “concerning ade*641quate and appropriate psychiatric care or safe and humane living conditions for persons institutionalized because of handicap or concerning payment of fair compensation to patients who perform work,” 41 Fed. Reg. 29548, 29559, the Solicitor General explains:

“This conclusion of course was consistent with the fact that, as relevant here, Section 504 is essentially concerned only with discrimination in the relative treatment of handicapped and nonhandicapped persons and does not confer any absolute right to receive particular services or benefits under federally assisted programs.” Brief for Petitioner 40, n. 33.

See also 48 Fed. Reg. 30846 (1983) (“Section 504 is in essence an equal treatment, non-discrimination standard”).28

The Final Rules, however, impose just the sort of absolute obligation on state agencies that the Secretary had previously disavowed. The services state agencies are required to make available to handicapped infants are in no way tied to the level of services provided to similarly situated nonhandicapped infants. Instead, they constitute an “absolute right to receive particular services or benefits” under a federally assisted program. Even if a state agency were scrupulously impartial as between the protection it offered handicapped and nonhandicapped infants, it could still be denied federal funding for failing to carry out the Secretary’s mission with sufficient zeal.

It is no answer to state, as does the Secretary, that these regulations are a necessary “ ‘metho[d] ... to give reasonable'assurance’ of compliance.” 49 Fed. Reg. 1627 (1984) (quoting 45 CFR § 80.4(b), which requires state agencies to *642report on their compliance with Title VI). For while the Secretary can require state agencies to document their own compliance with § 504, nothing in that provision authorizes him to commandeer state agencies to enforce compliance by other recipients of federal funds (in this instance, hospitals). State child protective services agencies are not field offices of the HHS bureaucracy, and they may not be conscripted against their will as the foot soldiers in a federal crusade.29 As we stated in Alexander v. Choate, 469 U. S., at 307, “nothing in the pre- or post-1973 legislative discussion of § 504 suggests that Congress desired to make major inroads on the States’ longstanding discretion to choose the proper mix” of services provided by state agencies.

VIII

Section 504 authorizes any head of an Executive Branch agency — regardless of his agency’s mission or expertise — to promulgate regulations prohibiting discrimination against the handicapped. See S. Rep. No. 93-1297, pp. 39-40 (1974).30 As a result of this rulemaking authority, the Secretary of *643HHS has “substantial leeway to explore areas in which discrimination against the handicapped pos[es] particularly significant problems and to devise regulations to prohibit such discrimination.” Alexander v. Choate, 469 U. S., at 304, n. 24.

Even according the greatest respect to the Secretary’s action, however, deference cannot fill the lack of an evidentiary foundation on which the Final Rules must rest. The Secretary’s basis for federal intervention is perceived discrimination against handicapped infants in violation of § 504, and yet the Secretary has pointed to no evidence that such discrimination occurs. Neither the fact that regulators generally may rely on generic information in a particular field or comparable experience gained in other fields, nor the fact that regulations may be imposed for preventative or prophylactic reasons, can substitute for evidence supporting the Secretary’s own chosen rationale. For the principle of agency accountability recited earlier means that “an agency’s action must be upheld, if at all, on the basis articulated by the agency itself.” Motor Vehicle Mfrs. Assn. v. State Farm Mut. Automobile Ins. Co., 463 U. S., at 50 (citations omitted).31

The need for a proper evidentiary basis for agency action is especially acute in this case because Congress has failed to indicate, either in the statute or in the legislative history, that it envisioned federal superintendence of treatment decisions traditionally entrusted to state governance. “[W]e must assume that the implications and limitations of our federal system constitute a major premise of all congressional legislation, though not repeatedly recited therein.” United States v. Gambling Devices, 346 U. S. 441, 450 (1953) (opin*644ion of Jackson, J.).32 Congress therefore “will not be deemed to have significantly changed the federal-state balance,” United States v. Bass, 404 U. S. 336, 349 (1971) — or to have authorized its delegates to do so — “unless otherwise the purpose of the Act would be defeated,” FTC v. Bunte Bros., Inc., 312 U. S. 349, 351 (1941).33 Although the nondiscrimi*645nation mandate of § 504 is cast in language sufficiently broad to suggest that the question is “not one of authority, but of its appropriate exercise[,] [t]he propriety of the exertion of the authority must be tested by its relation to the purpose of the [statutory] grant and with suitable regard to the principle that whenever the federal power is exerted within what would otherwise be the domain of state power, the justification of the exercise of the federal power must clearly appear.” Florida v. United States, 282 U. S. 194, 211-212 (1931). Accord, Chicago, M., St. P. & P. R. Co. v. Illinois, 355 U. S. 300, 306 (1958). That is, “it must appear that there are findings, supported by evidence, of the essential facts . . . which would justify [the Secretary’s] conclusion.” Florida v. United States, 282 U. S., at 212. The administrative record does not contain the reasoning and evidence that is necessary to sustain federal intervention into a historically state-administered decisional process that appears — for lack of any evidence to the contrary — to be functioning in full compliance with § 504.

The history of these regulations exposes the inappropriateness of the extraordinary deference — virtually a carte blanche — requested by the Government. The Secretary’s *646present reading of § 504 has evolved only after previous, patently erroneous interpretations had been found wanting.34 The checkered history of these regulations began in 1982, when the Department notified hospitals that they would violate § 504 if they “allow[ed] an infant” to remain in their care after “the infant’s parents or guardian [had withheld consent to] treatment or nourishment discriminatorily.” 47 Fed. Reg. 26027. By the time the Proposed Rules were announced one year later, the Secretary had abandoned that construction. But the Department substituted the equally untenable view that “the basic provision of nourishment, fluids, and routine nursing care” was “not an option for medical judgment” and that “[t]he decision to forego medical treatment of a correctable life-threatening defect because an infant also suffers from a permanent irremediable handicap that is not life-threatening, such as mental retardation, is a violation of Section 504,” insinuating by omission that lack of parental consent did not alter the hospital’s obligation to provide corrective surgery. 48 Fed. Reg. 30852, 30847 (1983). Although the preamble to the Final Rules corrects the prior erroneous signals from the Department that § 504 authorizes it to override parental decisions and to save the lives of handicapped infants, it persists in advocating federal regulation on the basis of treatment denials precipitated by refusals of parental consent and on the ground that its experience with the Baby Doe hotline has demonstrated that “the assumption that handicapped infants will receive medically beneficial treatment is not always justified.” 49 Fed. Reg. 1646 (1984).

This response, together with its previous remarks, makes irresistible the inference that the Department regards its *647mission as one principally concerned with the quality of medical care for handicapped infants rather than with the implementation of §504. We could not quarrel with a decision by the Department to concentrate its finite compliance resources on instances of life-threatening discrimination rather than instances in which merely elective care has been withheld. Cf. Heckler v. Chaney, 470 U. S. 821 (1985). But nothing in the statute authorizes the Secretary to dispense with the law’s focus on discrimination and instead to employ federal resources to save the lives of handicapped newborns, without regard to whether they are victims of discrimination by recipients of federal funds or not. Section 504 does not authorize the Secretary to give unsolicited advice either to parents, to hospitals, or to state officials who are faced with difficult treatment decisions concerning handicapped children. We may assume that the “qualified professionals” employed by the Secretary may make valuable contributions in particular cases, but neither that assumption nor the sincere conviction that an immediate “on-site investigation” is “necessary to protect the life or health of a handicapped individual” can enlarge the statutory powers of the Secretary.

The administrative record demonstrates that the Secretary has asserted the authority to conduct on-site investigations, to inspect hospital records, and to participate in the decisional process in emergency cases in which there was no colorable basis for believing that a violation of § 504 had occurred or was about to occur. The District Court and the Court of Appeals correctly held that these investigative actions were not authorized by the statute and that the regulations which purport to authorize a continuation of them are invalid.

The judgment of the Court of Appeals is affirmed.

It is so ordered.

Justice White,

dissenting.

Section 504 of the Rehabilitation Act of 1973 forbids discrimination solely on the basis of handicap in programs or activities receiving federal financial assistance. The issue before us is whether the Secretary of Health and Human Services has any authority under the Act to regulate medical treatment decisions concerning handicapped newborn infants. Relying on its prior decision in United States v. University Hospital, 729 F. 2d 144 (CA2 1984), the Court of Appeals held that the Secretary was without power in this respect and affirmed a decision of the District Court that § 504 does not extend so far and that the Secretary may not regulate such decisions in any manner.

Although it is my view that we granted certiorari to address this issue, the plurality avoids it by first erroneously reading the decision below as enjoining only the enforcement of specific regulations and by then affirming on the basis that the promulgation of the regulations did not satisfy established principles of administrative law, a matter that the Court of Appeals had no occasion to, and did not, discuss. With all due respect, I dissent.

I

The plurality’s initial and fundamental error is its statement that the only question presented here is the specific question whether the four mandatory provisions of the Final Rules issued by the Secretary are authorized by § 504. This conclusion misconstrues the opinion and judgment of the Court of Appeals. The plurality concedes that the District Court’s judgment on its face did not stop with enjoining the *649enforcement of the final regulations. Ante, at 625-626, n. 11. In fact, the District Court permanently enjoined the Secretary from implementing the final regulations and also from “continuing or undertaking any other actions to investigate or regulate treatment decisions involving impaired newborn infants taken under authority of Section 504, including pending investigation and other enforcement actions.” App. to Pet. for Cert. 51a-52a. This broad injunction ousted the Secretary from the field entirely and granted the precise relief sought by the complaint, which was filed after University Hospital and which sought to take full advantage of that decision.1 The Court of Appeals affirmed and in no way modified the injunction that the District Court had entered. In doing so, the Court of Appeals relied on its previous determination in University Hospital that the Secretary had no statutory authority to regulate medical treatment decisions regarding newborn infants. See App. to Pet. for Cert. 2a-3a.2

*650It is true that the regulations themselves were invalidated and their enforcement enjoined. This result, however, was directly compelled by the University Hospital conclusion that the Secretary was without power to issue any regulations whatsoever that dealt with infants’ medical care, and it did not comprise the whole relief awarded by the District Court and affirmed by the Court of Appeals. I thus see no justification for the plurality’s distortion of the Court of Appeals’ affirmance of the District Court’s all-inclusive injunction, which, like University Hospital, now represents the law in the Second Circuit.3 We should resolve the threshold statutory question that this case and University Hospital clearly pose — namely, whether the Secretary has any authority at all under the Act to regulate medical care decisions with respect to the handicapped newborn.4

II

Section 504 of the Act, which was construed in University Hospital, provides:

“No otherwise qualified handicapped individual in the United States, as defined in section 706(7) of this title, *651shall, solely by reason of his handicap, be excluded from the participation in, be denied the benefits of, or be subjected to discrimination under any program or activity receiving Federal financial assistance.” 29 U. S. C. §794.

After determining that § 706(7), which defines handicapped persons, is not limited to adults and includes the newborn, the Court of Appeals in University Hospital construed the “otherwise qualified” language of § 504 to limit the reach of the section to situations in which the handicap is “unrelated to, and thus improper to consideration of, the services in question.” 729 F. 2d, at 156.5 This, concluded the Court of Appeals, would exclude most handicapped newborns because their handicaps are not normally irrelevant to the need for medical services. Furthermore, the Court of Appeals thought that the “otherwise qualified” limitation should not be applied in the “comparatively fluid context of medical treatment decisions” because “[wjhere the handicapping condition is related to the condition(s) to be treated, it will *652rarely, if ever, be possible to say with certainty that a particular decision was ‘discriminatory.’” Id., at 156-157.

Having identified these perceived incongruities between the language of § 504 and the potential regulation of medical decisions regarding handicapped newborns, the Court of Appeals concluded that “[bjefore ruling that congress intended to spawn this type of litigation under section 504, we would want more proof than is apparent from the face of the statute.” Id., at 157. Thus, the Court of Appeals turned to the legislative history, where it again found nothing to persuade it that Congress intended § 504 to apply to medical treatment of handicapped infants and hence to enter a field so traditionally occupied by the States. Neither did it consider the current administrative interpretation of §504 to be a longstanding agency construction calling for judicial deference. In the Court of Appeals’ view, therefore, the section was inapplicable to medical treatment decisions regarding the newborn absent some further indication of congressional intent.

I disagree with this conclusion, which the Court of Appeals adhered to in the case before us now. Looking first at the language of the statute, I agree with the Court of Appeals’ preliminary conclusion that handicapped newborns are handicapped individuals covered by the Act. There is no reason for importing an age limitation into the statutory definition, and this Court has previously stated that “§504 protects handicapped persons of all ages from discrimination in a variety of programs and activities receiving federal financial assistance.” Smith v. Robinson, 468 U. S. 992, 1016-1017 (1984).6 This leaves the critical question whether a handi*653capped infant can ever be “otherwise qualified” for medical treatment and hence possibly subjected to unlawful discrimination when he or she is denied such treatment.7

*654It may well be that our prior consideration of this language has implied that the Court of Appeals’ construction is correct. In Southeastern Community College v. Davis, 442 U. S. 397, 406 (1979), we held that “[a]n otherwise qualified person is one who is able to meet all of a program’s requirements in spite of his handicap.” This formulation may be read as implying that where a handicapped person meets all of the requirements normally necessary to receive a program’s benefits regardless of his or her handicap, he or she is otherwise qualified because that handicap does not interfere with and is thus irrelevant to his or her qualification for the program. Thus, the Court of Appeals’ view — that refusing treatment that is called for only because of the handicapping condition cannot constitute discrimination on the basis of handicap since there will be no similarly situated nonhandicapped newborn, i. e., one who needs the same treatment — draws support from our holding in Davis since it turns on the same underlying perception that discrimination occurs only when the handicapping condition is irrelevant to the qualification for the program.

*655Even under the Court of Appeals’ interpretation of “otherwise qualified,” however, it does not follow that § 504 may never apply to medical treatment decisions for the newborn. An esophageal obstruction, for example, would not be part and parcel of the handicap of a baby suffering from Down’s syndrome, and the infant would benefit from and is thus otherwise qualified for having the obstruction removed in spite of the handicap. In this case, the treatment is completely unrelated to the baby’s handicapping condition. If an otherwise normal child would be given the identical treatment, so should the handicapped child if discrimination on the basis of the handicap is to be avoided.8

It would not be difficult to multiply examples like this. And even if it is true that in the great majority of cases the handicap itself will constitute the need for treatment, I doubt that this consideration or any other mentioned by the Court of Appeals justifies the wholesale conclusion that § 504 never applies to newborn infants with handicaps. That some or most failures to treat may not fall within § 504, that discerning which failures to treat are discriminatory may be difficult, and that applying § 504 in this area may intrude into the traditional functions of the State do not support the categori*656cal conclusion that the section may never be applied to medical decisions about handicapped infants. And surely the absence in the legislative history of any consideration of handicapped newborns does not itself narrow the reach of the statutory language. See Jefferson County Pharmaceutical Assn. v. Abbott Laboratories, 460 U. S. 150, 159-162, and n. 18 (1983). Furthermore, the broad remedial purpose of the section would be undermined by excluding handicapped infants from its coverage; and if, as the plurality indicates, ante, at 642-643, the Secretary has substantial leeway to explore areas in which discrimination against the handicapped poses serious problems and to devise regulations to prohibit the discrimination, it is appropriate to take note of the Secretary’s present view that § 504 properly extends to the subject matter at issue here. Thus, I believe that the Court of Appeals in University Hospital incorrectly concluded that § 504 may never apply to medical treatment decisions concerning handicapped newborn infants. Where a decision regarding medical treatment for a handicapped newborn properly falls within the statutory provision, it should be subject to the constraints set forth in §504. Consequently, I would reverse the judgment below.

Ill

Having determined that the stated basis for the Court of Appeals’ holding in University Hospital was incorrect and that the decision below cannot be supported by University Hospital’s blanket prohibition, I would remand the case to the Court of Appeals. Respondents have, as the plurality’s opinion itself demonstrates, raised significant issues aside from the threshold statutory issue presented here. There are, for example, substantial questions regarding the scope of the Secretary’s statutory authority in this area and whether these particular regulations are consistent with the statute. I would decline to reach and decide these questions for the first time in this Court without the benefit of the *657lower courts’ deliberations.9 The plurality, however, has chosen to reach out and address one of those subsidiary issues. Because the plurality has resolved that issue in a manner that I find indefensible on its own terms, I too address it.

The plurality concludes that the four mandatory provisions of the final regulations are invalid because there is no “ ‘rational connection between the facts found and the choice made.’” Motor Vehicle Mfrs. Assn., Inc. v. State Farm Mutual Automobile Ins. Co., 463 U. S. 29, 43 (1983) (quoting Burlington Truck Lines, Inc. v. United States, 371 U. S. 156, 168 (1962)). The basis for this conclusion is the plurality’s perception that two and only two wholly discrete categories of decisions are the object of the final regulations: (1) decisions made by hospitals to treat or not treat where parental consent has been given and (2) decisions made by hospitals to refer or not to refer a case to the state child protective services agency where parental consent has been withheld.10 *658Since the Secretary has not specifically pointed to discriminatory actions that provably resulted from either of these two specific types of decisions, the plurality finds that the Secretary’s conclusion that discrimination is occurring is unsupported factually. The plurality’s characterization of the Secretary’s rationale, however, oversimplifies both the complexity of the situations to which the regulations are addressed and the reasoning of the Secretary.

First, the Secretary’s proof that treatment is in fact being withheld from handicapped infants is unquestioned by the plurality. It is therefore obvious that whoever is making them, decisions to withhold treatment from such infants are in fact being made. This basic understanding is critical to the Secretary’s further reasoning, and the discussion accompanying the proposed regulations clearly indicates that this was the Secretary’s starting point. See 48 Fed. Reg. 30847-30848 (1983). Proceeding with this factual understanding, the next question is whether such withholding of treatment constitutes prohibited discrimination under § 504 in some or all situations. It is at this point that the plurality errs. In the plurality’s view, only two narrow paradigmatic types of decisions were contemplated by the Secretary as potentially constituting discrimination in violation of the statute. See ante, at 628-629. The plurality does not explain, however, precisely what in the Secretary’s discussion gives rise to this distillation, and my reading of the explanation accompanying the regulations does not leave me with so limited a view of the Secretary’s concerns.

The studies cited by the Secretary in support of the regulations and other literature concerning medical treatment in this area generally portray a decisionmaking process in which the parents and the doctors and often other concerned persons as well are involved — although the parental decision to consent or not is obviously the critical one.11 Thus, the pa*659rental consent decision does not occur in a vacuum. In fact, the doctors (directly) and the hospital (indirectly) in most cases participate in the formulation of the final parental decision and in many cases substantially influence that decision.12 Consequently, discrimination against a handicapped infant may assume guises other than the outright refusal to treat once parental consent has been given. Discrimination may occur when a doctor encourages or fails to discourage a parental decision to refuse consent to treatment for a handicapped child when the doctor would discourage or actually oppose a parental decision to refuse consent to the same treatment for a nonhandicapped child. Or discrimination may occur when a doctor makes a discriminatory treatment recommendation that the parents simply follow. Alternatively, discrimination may result from a hospital’s explicit laissez-faire attitude about this type of discrimination on the part of doctors.

Contrary to the plurality’s constrained view of the Secretary’s justification for the regulations, the stated basis for those regulations reveals that the Secretary was cognizant of this more elusive discrimination. For example, the evidence cited most extensively by the Secretary in his initial proposal of these regulations was a study of attitudes of practicing and teaching pediatricians and pediatric surgeons. See 48 Fed. Reg. 30848 (1983) (citing Shaw, Randolph, & Manard, Ethical Issues in Pediatric Surgery: A National Survey of Pediatricians and Pediatric Surgeons, 60 Pediatrics 588 (1977)). This study indicated that a substantial number of these doctors (76.8% of pediatric surgeons and 49.5% of pedi*660atricians) would “acquiesce in parents’ decision to refuse consent for surgery in a newborn with intestinal atresia if the infant also had . . . Down’s syndrome.” Id., at 590. It also indicated that a substantial minority (23.6% of pediatric surgeons and 13.2% of pediatricians) would in fact encourage parents to refuse consent to surgery in this situation and that only a small minority (3.4% of pediatric surgeons and 15.8% of pediatricians) would attempt to get a court order mandating surgery if the parents refused consent. In comparison, only a small minority (7.9% of pediatric surgeons and 2.6% of pediatricians) would acquiesce in parental refusal to treat intestinal atresia in an infant with no other anomaly. And a large majority (78.3% of pediatric surgeons and 88.4% of pediatricians) would try to get a court order directing surgery if parental consent were withheld for treatment of a treatable malignant tumor. The Secretary thus recognized that there was evidence that doctors would act differently in terms of attempts to affect or override parental decisions depending on whether the infant was handicapped.

Based on this evidence, the Secretary conceded that “[t]he full extent of discriminatory and life-threatening practices toward handicapped infants is not yet known” but concluded “that for even a single infant to die due to lack of an adequate notice and complaint procedure is unacceptable.” 48 Fed. Reg. 30847 (1983). Thus, the Secretary promulgated the regulations at issue here. These regulations, in relevant part, require that a notice of the federal policies against discrimination on the basis of handicap be posted in a place where a hospital’s health care professionals will see it. This requirement is, as the Secretary concluded, “[consistent with the Department’s intent to target the notice to nurses and other health care professionals.” App. 25. The notice requirement, therefore, may reasonably be read as aimed at fostering an awareness by health care professionals of their responsibility not to act in a discriminatory manner with respect to medical treatment decisions for handicapped infants. *661The second requirement of the regulations, that state agencies provide mechanisms for requiring and reporting medical neglect of handicapped children, is also consistent with the Secretary’s focus on discrimination in the form of discriminatory reporting.13

I therefore perceive a rational connection between the facts found by the Secretary and the regulatory choice made. The Secretary identified an existing practice that there was reason to believe resulted from discrimination on the basis of handicap. Given this finding, the amorphous nature of much of the possible discrimination, the Secretary’s profession that the regulations are appropriate no matter how limited the problem,14 and the focus of the regulations on loci where unlawful discrimination seems most likely to occur and on persons likely to be responsible for it, I conclude that these regulations are not arbitrary and capricious and that the Court errs in striking them down on that basis. Although the Secretary’s path here may be marked with “‘less than ideal clarity,’ ” we will uphold such a decision “ ‘if the agency’s path may reasonably be discerned.’” Motor Vehicles Mfrs. Assn., 463 U. S., at 43 (quoting Bowman Transportation, Inc. v. Arkansas-Best Freight System, Inc., 419 U. S. 281, 286 (1974)).

The plurality also objects to the regulations’ requirement concerning the state protective agencies’ reporting proce*662dures on another ground. Specifically, the plurality finds that this requirement is in fact a substantive prescription rather than a prohibition of discrimination. The plurality bases this conclusion on the fact that the regulation sets forth specific procedures that must be adopted by state agencies.

The plurality’s conclusion disregards the Secretary’s explanation for this requirement. In the preamble to the proposed regulations, the Secretary explicitly stated:

“The Department has determined that under every state’s law, failure of parents to provide necessary, medically indicated care to a child is either explicitly cited as grounds for action by the state to compel treatment or is implicitly covered by the state statute. These state statutes also provide for appropriate administrative and judicial enforcement authorities to prevent such instances of medical neglect, including requirements that medical personnel report suspected cases to the state child protective services agency, agency access to medical files, immediate investigations and authority to compel treatment.” 48 Fed. Reg. 30848 (1983).

This finding was repeated in the statement accompanying the final regulations:

“Although there are some variations among state child protective statutes, all have the following basic elements: a requirement that health care providers report suspected cases of child abuse or neglect, including medical neglect; a mechanism for timely receipt of such reports; a process for administrative inquiry and investigation to determine the facts; and the authority and responsibility to seek an appropriate court order to remedy the apparent abuse and neglect, if it is found to exist.” 49 Fed. Reg. 1627 (1984).

The regulations, in turn, require that the State provide these same services with respect to medical neglect of handicapped infants. See 45 CFR §84.55(c) (1985). The only *663additional requirements imposed by the regulations involve provisions enabling the Department itself to review for compliance with the nondiscrimination requirements. Consequently, the regulations simply track the existing state procedures found to exist by the Secretary, requiring that funded state agencies provide those same procedures for handicapped children. The fact that the regulations specify the procedures that are necessary to ensure an absence of discrimination and do not instead speak in “nondiscrimination” terms is irrelevant. The substance of the requirement is nondiscrimination. The plurality’s conclusion in this regard, however, apparently rests on a determination that implementation of a nondiscrimination mandate may be accomplished in only one form — even if the same result may be accomplished by another route. See ante, at 640, n. 26. I would not elevate regulatory form over statutory substance in this manner. In sum, the plurality’s determination that the regulations were inadequately supported and explained as a matter of administrative law does not withstand examination of the Secretary’s discussion of the underlying problem and of the contours of the regulations themselves.

IV

My disagreement with the plurality in this case does not end here, however. For even under its chosen rationale, I find its ultimate conclusion dubious. Having assiduously restricted its discussion to the validity of the regulations only, the plurality ends up concluding expansively that not only the regulations but also other investigations taken by the Secretary independent of the regulations are invalid. Thus, the Court apparently enjoins the Secretary’s on-site investigations as well as “the regulations which purport to authorize a continuation of them.” Ante, at 647. And the plurality rests this action on the conclusion that the lower courts “correctly held that these investigative actions were not authorized by the statute.” Ibid.

*664I am at a loss to understand the plurality’s reasoning in this respect. In construing the judgment below, the plurality appears to conclude that, although the injunction entered by the District Court and affirmed by the Court of Appeals did not purport to prohibit all actions by the Secretary under the statute, the injunction did in fact extend beyond merely these particular regulations. Thus, the plurality indicates that the judgment below applied as well to actions that “resemble,” “parallel,” or are “along [the] lines [of]” the regulations. Ante, at 625-626, n. 11. The plurality further defines what actions it believes the Court of Appeals and District Court contemplated: “[T]he injunction forbids continuation or initiation of regulatory and investigative activity directed at instances in which parents have refused consent to treatment and, if the Secretary were to undertake such action, efforts to seek compliance with affirmative requirements imposed on state child protective services agencies.” Ante, at 625, n. 11.

Aside from the fact that I see absolutely nothing in either the District Court’s or the Court of Appeals’ judgment that would support a constrained reading of the broadly phrased relief awarded by the District Court and affirmed without modification by the Court of Appeals,151 have some doubt as to how different the Court’s holding today is from a holding that § 504 gives HHS no authority whatsoever over decisions to treat handicapped infants. The plurality’s lack of coherence on this crucial point raises substantial doubts as to the reach of the holding and as to the basis for that holding.

Finally, I am puzzled as to how and why the plurality’s determination that the regulations are invalid because they are arbitrary and capricious extends to other actions not taken under the regulations. The plurality apparently would enjoin all enforcement actions by the Secretary in situations in which parents have refused to consent to treatment. See ante, at 625-626, n. 11. Yet it is not clear to me that the *665plurality’s basis for invalidating these regulations would extend to all such situations. I do not see, for example, why the plurality’s finding that the Secretary did not adequately support his conclusion that failures to report refusals to treat likely result from discrimination means that such a conclusion will never be justified. The Secretary might be able to prove that a particular hospital generally fails to report nontreatment of handicapped babies for a specific treatment where it reports nontreatment of nonhandicapped babies for the same treatment. In essence, a determination that these regulations were inadequately supported factually would not seem to be properly extended beyond actions taken pursuant to these regulations: The fact that the Secretary has not adequately justified generalized action under the regulations should not mean that individualized action in appropriate circumstances is precluded.

V

In sum, the plurality today mischaracterizes the judgment below and, based on that mischaracterization, is sidetracked from the straightforward issue of statutory construction that this case presents. The plurality incorrectly resolves an issue that was not fully addressed by the parties, gives no guidance to the Secretary or the other parties as to the proper construction of the governing statute, and fails adequately to explain the precise scope of the holding or how that holding is supported under the plurality’s chosen rationale. From this misguided effort, I dissent.

Justice O’Connor,

dissenting.

I fully agree with Justice White’s conclusion that the only question properly before us is whether the Court of Appeals correctly concluded that the Secretary has no power under 29 U. S. C. § 794 to regulate medical treatment decisions concerning handicapped newborn infants. I also agree that application of established principles of statutory construction and of the appropriate standard for judicial review *666of agency action leads inescapably to the conclusion that the Secretary has the authority to regulate in this area. Because, however, I see no need at this juncture to address the details of the regulations or to assess whether they are sufficiently rational to survive review under 5 U. S. C. §706 (2)(A), I join only parts I, II, IV, and V of Justice White’s dissent.

FINDINGS OF FACT AND CONCLUSIONS OF LAW

HILTON, District Judge.

This case was tried before the court, and upon the evidence presented and argument of counsel, the court makes the following Findings of Fact and Conclusions of Law.

Findings of Fact

1. Plaintiff Hospital is a general acute care hospital located in Virginia that is licensed to provide diagnosis, treatment, and medical and nursing services to the public as . provided by Virginia law. Among other facilities, the Hospital has a Pediatric Intensive Care Department and an Emergency Department.

2. The Hospital is a recipient of federal and state funds including those from Medicare and Medicaid and is a “participating hospital” pursuant to 42 U.S.C. § 1395cc.

3. The Hospital and its staff (including emergency doctors, pediatricians, neonatologists and pediatric intensivists) treat sick children on a daily basis.

4. Defendant Ms. H, a citizen of the Commonwealth of Virginia, is the biological mother of Baby K, an infant girl born by Caesare *1025 an section at the Hospital on October 13, 1992. Baby K was born with anencephaly.

5. Anencephaly is a congenital defect in which the brain stem is present but the cerebral cortex is rudimentary or absent. There is no treatment that will cure, correct, or ameliorate anencephaly. Baby K is permanently unconscious and cannot hear or see. Lacking a cerebral function, Baby K does not feel pain. Baby K has brain stem functions primarily limited to reflexive actions such as feeding reflexes (rooting, sucking, swallowing), respiratory reflexes (breathing, coughing), and reflexive responses to sound or touch. Baby K has a normal heart rate, blood pressure, liver function, digestion, kidney function, and bladder function and has gained weight since her birth. Most anencephalic infants die within days of birth.

6. Baby K was diagnosed prenatally as being anencephalic. Despite the counselling of her obstetrician and neonatologist that she terminate her pregnancy, Ms. H refused to have her unborn child aborted.

7. A Virginia court of competent jurisdiction has found defendant Mr. K, a citizen of the Commonwealth of Virginia, to be Baby K’s biological father.

8. Ms. H and Mr. K have never been married.

9. Since Baby K’s birth, Mr. K has, at most, been only distantly involved in matters relating to the infant. Neither the Hospital nor Ms. H ever sought Mr. K’s opinion or consent in providing medical treatment to Baby K.

10. Because Baby K had difficulty breathing immediately upon birth, Hospital physicians provided her with mechanical ventilator treatment to allow her to breathe.

11. Within days of Baby K’s birth, Hospital medical personnel urged Ms. H to permit a “Do Not Resuscitate Order” for Baby K that would discontinue ventilator treatment. Her physicians told her that no treatment existed for Baby K’s anencephalic condition, no therapeutic or palliative purpose was served by the treatment, and that ventilator care was medically unnecessary and inappropriate. Despite this pressure, Ms. H eontinued to request ventilator treatment for her child.

12. Because of Ms. H’s continued insistence that Baby K receive ventilator treatment, her treating physicians requested the assistance of the Hospital’s “Ethics Committee” in overriding the mother’s wishes.

13. A three person Ethics Committee subcommittee, composed of a family practitioner, a psychiatrist, and a minister, met with physicians providing care to Baby K. On October 22, 1992, the group concluded that Baby K’s ventilator treatment should end because “such care is futile” and decided to “wait a reasonable time for the family to help the caregiver terminate aggressive therapy.” If the family refused to follow this advice, the committee recommended that the Hospital should “attempt to resolve this through our legal system.”

14. Ms. H subsequently rejected the committee’s recommendation. Before pursuing legal action to override Ms. H’s position, the Hospital decided to transfer the infant to another health care facility.

15. Baby K was transferred to a nursing home (“Nursing Home”) in Virginia on November 30, 1992 during a period when she was not experiencing respiratory distress and thus did not need ventilator treatment. A condition of the transfer was that the Hospital agreed to take the infant back if Baby K again developed respiratory distress to receive ventilator treatment which was unavailable at the Nursing Home. Ms. H agreed to this transfer.

16. Baby K returned to the Hospital on January 15, 1993 after experiencing respiratory distress to receive ventilator treatment. Hospital officials again attempted to persuade Ms. H to discontinue ventilator treatment for her child. Ms. H again refused. After Baby K could breathe on her own, she was transferred back to the Nursing Home on February 12, 1993.

17. Baby K again experienced breathing difficulties on March 3, 1993 and returned to the Hospital to receive ventilator treatment.

18. On March 15, 1993, Baby K received a tracheotomy, a procedure in which a *1026 breathing tube is surgically implanted in her windpipe, to facilitate ventilator treatment. Ms. H agreed to this operation.

19. After no longer requiring ventilator treatment, Baby K was transferred back to the Nursing Home on April 13, 1993 where she continues to live.

20. Baby K will almost certainly continue to have episodes of respiratory distress in the future. In the absence of ventilator treatment during these episodes, she would suffer serious impairment of her bodily functions and soon die.

21. Ms. H visits Baby K daily. The mother opposes the discontinuation of ventilator treatment when Baby K experiences respiratory distress because she believes that all human life has value, including her anencephalic daughter’s life. Ms. H has a firm Christian faith that all life should be protected. She believes that God will work a miracle if that is his will. Otherwise, Ms. H believes, God, and not other humans, should decide the moment of her daughter’s death. As Baby K’s mother and as the only parent who has participated in the infant’s care, Ms. H believes that she has the right to decide what is in her child’s best interests.

22. On the Hospital’s motion, a guardian ad litem to represent Baby K was appointed pursuant to Virginia Code § 8.01-9.

23. Both the guardian ad litem and Mr. K share the Hospital’s position that ventilator treatment should be withheld from Baby K when she experiences respiratory distress.

24. The Hospital has stipulated that it is not proposing to deny ventilator treatment to Baby K because of any lack of adequate resources or any inability of Ms. H to pay for the treatment.

Conclusions of Law

Pursuant to the Declaratory Judgment Act, 28 U.S.C. § 2201, the Hospital has sought declaratory and injunctive relief under four federal statutes and one Virginia statute: the Emergency Medical Treatment and Active Labor Act, 42 U.S.C. § 1395dd; the Rehabilitation Act of 1973, 29 U.S.C. § 794; the Americans with Disabilities Act of 1990, 42 U.S.C. § 12101 et seq.; the Child Abuse Amendments of 1984, 42 U.S.C. § 5102 et seq.; and the Virginia Medical Malpractice Act, Va.Code § 8.01-581.1 et seq. This court has federal question jurisdiction under the four federal statutes and supplemental jurisdiction regarding the Virginia statute. 28 U.S.C. §§ 1331, 1367.

I. Emergency Medical Treatment and Active Labor Act

Plaintiff seeks a declaration that its refusal to provide Baby K with life-supporting medical care would not transgress the Emergency Medical Treatment and Active Labor Act, 42 U.S.C. § 1395dd (“EMTALA”). EMTALA requires that participating hospitals provide stabilizing medical treatment to any person who comes to an emergency department in an “emergency medical condition” when treatment is requested on that person’s behalf. An “emergency medical condition” is defined in the statute as “acute symptoms of sufficient severity ... such that the absence of immediate medical attention could reasonably be expected to result in ... serious impairment to bodily functions, or serious dysfunction of any bodily organ or part.” 42 U.S.C. § 1395dd(e)(1)(A). “Stabilizing” medical treatment is defined as “such medical treatment of the condition as may be necessary to assure, within reasonable medical probability, that no material deterioration of the condition” will result. Id. § 1395dd(e)(3)(A). The statute’s legislative history includes a position paper by the American College of Emergency Physicians stating that “stabilization” should include “[establishing and assuring an adequate airway and adequate ventilation.” H.R.Rep. No. 241 (Pt. 3), 99th Cong., 1st Sess. 26 (1985).

The Hospital admits that Baby K would meet these criteria if she is brought to the Hospital while experiencing breathing difficulty. As stated in the Hospital’s complaint, when Baby K is in respiratory distress, that condition is “such that the absence of immediate medical attention could reasonably be expected to cause serious impairment to her bodily functions” — i.e., her breathing difficulties constitute an. “emergency medical condition.” The Hospital also concedes in its com *1027 plaint that ventilator treatment is required in such circumstances to assure “that no material deterioration of Baby K’s condition is likely to occur” — ie., a ventilator is necessary to “stabilize” the baby’s condition. These admissions establish that the Hospital would be liable under EMTALA if Baby K arrived there in respiratory distress (or some other emergency medical condition) and the Hospital failed to provide mechanical ventilation (or some other medical treatment) necessary to stabilize her acute condition.

The Hospital would also have an obligation to continue to provide stabilizing medical treatment to Baby K even if she were admitted to the pediatric intensive care unit or other unit of the Hospital and to provide the treatment until she could be transferred back to the Nursing Home or to another facility willing to accept her. See Thornton v. Southwest Detroit Hosp., 895 F.2d 1131, 1135 (6th Cir.1990) (“emergency care does not always stop when a patient is wheeled from the emergency room into the main hospital”); McIntyre v. Schick, 795 F.Supp. 777, 781 (E.D.Va.1992) (rationale behind “anti-dumping statute is not based upon the door of the hospital through which a patient enters, but rather upon the notion of proper medical care for those persons suffering medical emergencies, whenever such emergencies occur at a participating hospital. Indeed, it is. a ridiculous distinction, one which places form over substance, to state that the care a patient receives depends on the door through which the patient walks”).

Despite EMTALA’s clear requirements and in the face of the Hospital’s admissions, the Hospital seeks an exemption from the statute for instances in which the treatment at issue is deemed “futile” or “inhumane” by the hospital physicians. The plain language of the statute requires stabilization of an emergency medical condition. The statute does not admit of any “futility” or “inhumanity” exceptions. Any argument to the contrary should be directed to the U.S. Congress, not to the Federal Judiciary. Cf. Baber v. Hospital Corporation of America, 977 F.2d 872, 878 (4th Cir.1992) (rejecting argument that EMTALA provides private cause of action against physicians on grounds that “it is not our role to rewrite legislation passed by Congress. When a statute is clear and unambiguous, we must apply its terms as written instead of varying its terms to accommodate a perceived legislative intent”).

Even if EMTALA contained the exceptions advanced by the Hospital, these exceptions would not apply here. The use of a mechanical ventilator to assist breathing is not “futile” or “inhumane” in relieving the acute symptoms of respiratory difficulty which is the emergency medical condition that must be treated under EMTALA. To hold otherwise would allow hospitals to deny emergency treatment to numerous classes of patients, such as accident victims who have terminal cancer or AIDS, on the grounds that they eventually will die anyway from those diseases and that emergency care for them would therefore be “futile.”

II. Rehabilitation Act

Section 504 of the Rehabilitation Act prohibits discrimination against an “otherwise qualified” handicapped individual, solely by reason of his or her handicap, under any program or activity receiving federal financial assistance. Hospitals such as plaintiff that accept Medicare and Medicaid funding are subject to the Act. United States v. Baylor Univ. Medical Center, 736 F.2d 1039, 1049 (5th Cir.1984), cert. denied, 469 U.S. 1189, 105 S.Ct. 958, 83 L.Ed.2d 964 (1985). Baby K is a “handicapped” and “disabled” person within the meaning of the Rehabilitation Act of 1973. A “handicapped individual” under the Rehabilitation Act “includes an infant who is born with a congenital defect.” Bowen v. American Hospital Ass’n, 476 U.S. 610, 624, 106 S.Ct. 2101, 2110, 90 L.Ed.2d 584 (1986).

Section 504’s plain text spells out the necessary scope of inquiry: Is Baby K otherwise qualified to receive ventilator treatment and is ventilator treatment being threatened with being denied because of an unjustified consideration of her anencephalic handicap? The Hospital has admitted that the sole reason it wishes to withhold ventilator treatment for Baby K over her mother’s objections, is because of Baby K’s aneneephaly — her handicap and disability.

*1028 To evade this textual mandate, the Hospital relies on two cases which held that a hospital’s decision not to override the desire of the parents of babies with congenital defects to withhold treatment did not violate section 504. Johnson v. Thompson, 971 F.2d 1487, 1493 (10th Cir.1992), cert. denied, — U.S. -, 113 S.Ct. 1255, 122 L.Ed.2d 654 (1993); United States v. University Hospital, State U. of New York, 729 F.2d 144, 156-57 (2d Cir.1984). Because the parents in Johnson and University Hospital consented to the withholding of treatment, the two cases are factually distinguishable from this case. 1

When the Rehabilitation Act was passed in 1973, Congress intended that discrimination on the basis of a handicap be treated in the same manner that Title VI of the Civil Rights Act treats racial discrimination. University Hospital, 729 F.2d at 161-163 (Winter, J., dissenting). This analogy to race dispels any ambiguity about the extent to which Baby K has statutory rights not to be discriminated against on the basis of her handicap. It also shatters the Hospital’s contention that ventilator treatment should be withheld because Baby K’s recurring breathing troubles are intrinsically related to her handicap. No such distinction would be permissible within the context of racial discrimination. In addition, the Hospital was able to perform a tracheotomy on Baby K. This surgery was far more complicated than linking her to a ventilator to allow her to breathe. Cf. Bowen, 476 U.S. at 655, 106 S.Ct. at 2127 (“if an otherwise normal child would be given the identical treatment, so should the handicapped child”) (White, J., dictum in dissent). Just as an AIDS patient seeking ear surgery is “otherwise qualified” to receive treatment despite poor long term prospects of living, Baby K is “otherwise qualified” to receive ventilator treatment despite similarly dismal health prospects. Cf. Glanz v. Vernick, 750 F.Supp. 39, 45-46 (D.Mass.1990). Thus, the Hospital’s desire to withhold ventilator treatment from Baby K over her mother’s objections would violate the Rehabilitation Act.

III. Americans with Disabilities Act

Section 302 of the Americans with Disabilities Act (“ADA”) prohibits discrimination against disabled individuals by “public accommodations.” 42 U.S.C. § 12182. A “disability” is “a physical or mental impairment that substantially limits one or more of the major life activities” of an individual. 42 U.S.C. § 12102(2). This includes any physiological disorder or condition affecting the neurological system, musculoskeletal system, or sense organs, among others. 28 C.F.R. § 36.104 (definition of “physical or mental impairment”). Anencephaly is a disability, because it affects the baby’s neurological functioning, ability to walk, and ability to see or talk. “Public accommodation” is defined to include a “professional office of a health care provider, hospital, or other service establishment.” 42 U.S.C. § 12181(7). The Hospital is a public accommodation under the ADA. 28 C.F.R. § 36.104.

Section 302(a) of the ADA states a general rule of nondiscrimination against the disabled:

General rule. No individual shall be discriminated against on the basis of disability in the full and equal enjoyment of the goods, services, facilities, privileges, advantages, or accommodation of any place of public accommodations by any person who owns, leases (or leases to), or operates a place of public accommodation.

42 U.S.C. § 12182(a). In contrast to the Rehabilitation Act, the ADA does not require that a handicapped individual be “otherwise qualified” to receive the benefits of participation. Further, section 302(b)(1)(A) of the ADA states that “[i]t shall be discriminatory to subject an individual or class of individuals on the basis of a disability ... to a denial of the opportunity of the individual or class to participate in or benefit from the goods, services, facilities, privileges, advantages, or ac *1029 commodations of an entity.” 42 U.S.C. § 12182(b)(l)(A)(i).

The Hospital asks this court for authorization to deny the benefits of ventilator services to Baby K by reason of her aneneephaly. The Hospital’s claim is that it is “futile” to keep alive an anencephalie baby, even though the mother has requested such treatment. But the plain language of the ADA does not permit the denial of ventilator services that would keep alive an anencephalie baby when those life-saving services would otherwise be provided to a baby without disabilities at the parent’s request. The Hospital’s reasoning would lead to the denial of medical services to anencephalie babies as a class of disabled individuals. Such discrimination against a vulnerable population class is exactly what the American with Disabilities Act was enacted to prohibit. The Hospital would therefore violate the ADA if it were to withhold ventilator treatment from Baby K.

IV. Child Abuse Act

Plaintiff seeks a declaration that it may refuse to provide life-supporting medical care to Baby K without incurring liability under the Child Abuse Amendments of 1984, 42 U.S.C. § 5101 et seq. (“Child Abuse Act”). This request for relief must be denied because the Hospital has failed to join a necessary party — the Virginia Child Protective Services.

There is no private right of action against a health care provider under the Child Abuse Act. Jensen v. Conrad, 570 F.Supp. 91, 113 (D.S.C.1983) (citing Perry v. Housing Authority of Charleston, 664 F.2d 1210, 1213 (4th Cir.1981)), aff'd, 747 F.2d 185 (4th Cir.1984), cert. denied, 470 U.S. 1052, 105 S.Ct. 1754, 84 L.Ed.2d 818 (1985). The Act only authorizes states which receive federal grants for child abuse and neglect programs to bring legal action through their child protective services agencies to prevent the medical neglect of disabled infants. 42 U.S.C. § 5106a(b)(10)(C); 45 C.F.R. § 1340.-15(c)(2)(iii).

Because the Virginia Child Protective Services has an interest in a declaratory judgment regarding the Child Abuse Act and is the only party that can enforce the Act, it is a necessary party. This court must have the sole enforcing authority party before it before considering the declaratory judgment issue. ARW Exploration Corp. v. Aguirre, 947 F.2d 450, 454 (10th Cir.1991). Without the Virginia Child Protective Services as a party, no actual controversy exists as to this issue as required by the Declaratory Judgment Act. 28 U.S.C. § 2201. Thus, the Hospital’s request for declaratory and injunctive relief under the Child Abuse Amendments must be denied. See White v. National Union Fire Ins. Co., 913 F.2d 165, 167-168 (4th Cir.1990).

V. Virginia Medical Malpractice Act

The Hospital seeks a declaration that its refusal to provide Baby K with ventilator treatment does not constitute malpractice under the Virginia Medical Malpractice Act, Va.Code § 8.01-581.1 et seq. Under the Declaratory Judgment Act, 28 U.S.C. § 2201, a federal court has discretion to assert jurisdiction to render a declaration. Mitcheson v. Harris, 955 F.2d 235 (4th Cir.1992); Wright & Miller, Federal Practice and Procedure § 2759 (1983). This discretion is related to the interest in having states decide questions of state law, an interest having jurisprudential roots in Erie R.R. Co. v. Tompkins, 304 U.S. 64, 58 S.Ct. 817, 82 L.Ed. 1188 (1938). Mitcheson, 955 F.2d at 237. In the Declaratory Judgment Act, Congress has afforded federal courts a freedom to consider the state interest in having state courts determine questions of state law. Id. at 238. Virginia courts have not addressed the question of the appropriate standard of care for anencephalie infants and whether an exception to the general standard of care applies to them. Besides the Malpractice Act’s general rule, Virginia’s legislature has also been silent on the issue. Moreover, the determination of the standard of care under Virginia’s Medical Malpractice Act involves a review panel appointed by the Chief Justice of the Virginia Supreme Court. This review panel mechanism is “so intimately bound up with the rights and obligations being asserted as to require their application in federal courts under the doctrine of Erie Railroad Co. v. Tompkins.” DiAntonio v. Northampton-Accomack Memorial Hospital, 628 F.2d 287, *1030 290 (4th Cir.1980). Because of the significant state interests manifested by this review process as well as the Commonwealth’s interest in resolving this contentious and unsettled social issue for itself, this court declines to “elbow its way” into Virginia medical malpractice standards. Cf. Mitcheson, 955 F.2d at 238 (quoting Pennhurst State School & Hosp. v. Halderman, 465 U.S. 89, 122 n. 32, 104 S.Ct. 900, 920 n. 32, 79 L.Ed.2d 67 (1984)).

VI. Constitutional and Common Law Issues

Baby K’s parents disagree over whether or not to continue medical treatment for her. Mr. K and Baby K’s guardian ad litem join the Hospital in seeking the right to override the wishes of Ms. H, Baby K’s mother. Regardless of the questions of statutory interpretation presented in this case, Ms. H retains significant legal rights regarding her insistence that her daughter be kept alive with ventilator treatment. A parent has a constitutionally protected right to “bring up children” grounded in the Fourteenth Amendment’s due process clause. Meyer v. Nebraska, 262 U.S. 390, 399, 43 S.Ct. 625, 626, 67 L.Ed. 1042 (1923); Pierce v. Society of Sisters, 268 U.S. 510, 534-535, 45 S.Ct. 571, 573, 69 L.Ed. 1070 (1925). Parents have the “primary role” in the “nurture and upbringing of their children.” Wisconsin v. Yoder, 406 U.S. 205, 232, 92 S.Ct. 1526, 1541, 32 L.Ed.2d 15 (1972); Prince v. Massachusetts, 321 U.S. 158, 166, 64 S.Ct. 438, 442, 88 L.Ed. 645 (1944). Decisions for children can be based in the parent’s free exercise of religion, protected by the First Amendment. Pierce, 268 U.S. at 534-535, 45 S.Ct. at 573; Yoder, 406 U.S. at 234, 92 S.Ct. at 1542.

These constitutional principles extend to the right of parents to make medical treatment decisions for their minor children. Absent a finding of neglect' or abuse, parents retain plenary authority to seek medical care for their children, even when the decision might impinge on a liberty interest of the child. Parham v. J.R., 442 U.S. 584, 603-604, 99 S.Ct. 2493, 2504, 61 L.Ed.2d 101 (1979) (commitment of child to mental health hospital). Indeed, there is a “presumption that the parents act in the best interests of their child” because the “natural bonds of affection lead parents to act in the best interests of their children.” Id. at 602, 99 S.Ct. at 2504.

State law rights to make medical and surgical treatment decisions for a minor child are grounded in the common law and can also be inferred from state statutes. See Va.Code § 54.1-2969(B) (procedure governing consent to treatment of minors when parents are unavailable); Va.Code § 16.1-334(1) (right of emancipated minor to make her own medical care decisions without parental consent).

Based on Ms. H’s “natural bonds of affection,” Parham, 442 U.S. at 602, 99 S.Ct. at 2504, and the relative noninvolvement of Baby K’s biological father, the constitutional and common law presumption must be that Ms. H. is the appropriate decision maker. “[W]hen parents do not agree on the issue of termination of life support ... this Court must yield to the presumption in favor of life.” In re Jane Doe, A Minor, Civ. No. D-93064, mem. op. at 18 (Super.Ct. Fulton Co., Ga., October 17, 1991), aff'd, 262 Ga. 389, 418 S.E.2d 3 (1992).' This presumption arises from the explicit guarantees of a right to life in the United States Constitution, Amendments V and XIV, and the Virginia Constitution, Article 1, Sections 1 and 11.

The presumption in favor of life in this case is also based on Ms. H’s religious conviction that all life is sacred and must be protected, thus implicating her First Amendment rights. When an individual asserts “the Free Exercise Clause in conjunction with other constitutional protections, such as ... the right of parents,” only a clear and compelling governmental interest can justify a statute that interferes with the person’s religious convictions. Employment Div., Department of Human Resources of Oregon v. Smith, 494 U.S. 872, 881 n. 1, 110 S.Ct. 1595, 1601 n. 1, 108 L.Ed.2d 876 (1990); Yoder, 406 U.S. at 233, 92 S.Ct. at 1542.

The Hospital cannot establish any “clear and compelling” interest in this case. The Supreme Court has not decided whether the right to liberty encompasses a *1031 right to refuse medical treatment, often called a “right to die.” Cruzan v. Director, Missouri Dept. of Health, 497 U.S. 261, 277-279, 110 S.Ct. 2841, 2850-51, 111 L.Ed.2d 224 (refusing to decide this question). Parents have standing to assert the constitutional rights of their minor children. Eisenstadt v. Baird, 405 U.S. 438, 446 n. 6, 92 S.Ct. 1029, n. 6, 31 L.Ed.2d 349 (1972). When one parent asserts the child’s explicit constitutional right to life as the basis for continuing medical treatment and the other is asserting the nebulous liberty interest in refusing life-saving treatment on behalf of a minor child, the explicit right to life must prevail. See In re Jane Doe, supra.

Reflecting the constitutional principles of family autonomy and the presumption in favor of life, courts have generally scrutinized a family’s decision only where the family has sought to terminate or withhold medical treatment for an incompetent minor or incompetent adult. See, e.g., Cruzan, 497 U.S. at 270-75, 110 S.Ct. at 2847-49 (and cases cited therein). In a recent case in which a hospital sought to terminate life-supporting ventilation over the objections of the patient’s husband, a Minnesota state court refused to remove decisionmaking authority from the husband. In re Wanglie, No. PX-91-283 (Prob.Ct., Hennepin Co., Minn., June 28, 1991). Likewise, where parents disagreed over whether to continue life-supporting mechanical ventilation, nutrition, and hydration for a minor child in an irreversible stupor or coma, a Georgia state court gave effect to the decision of the parent opting in favor of life support. In re Jane Doe, supra. 2

At the very least, the Hospital must establish by clear and convincing evidence that Ms. H’s treatment decision should not be respected because it would constitute abuse or neglect of Baby K. This clear and convincing evidence standard has been adopted by numerous courts and was upheld by the Supreme Court in Cruzan in authorizing the withdrawal of life-supporting treatment from an incompetent patient. See Cruzan, 497 U.S. at 284-85, 110 S.Ct. at 2854-55. In this case, where the choice essentially devolves to a subjective determination as to the quality of Baby’s K’s life, it cannot be said that the continuation of Baby K’s life is so unreasonably harmful as to constitute child abuse or neglect.

For the foregoing reasons, the Hospital’s request for a declaratory judgment that the withholding of ventilator treatment from Baby K would not violate the Emergency Medical Treatment and Active Labor Act, the Rehabilitation Act of 1973, the Americans with Disabilities Act, the Child Abuse Amendments of 1984, and the Virginia Medical Malpractice Act should be DENIED. Under the Emergency Medical Treatment and Active Labor Act, the Rehabilitation Act of 1973, and the Americans with Disabilities Act, the Hospital is legally obligated to provide ventilator treatment to Baby K. The court makes no ruling as to any rights or obligations under the Child Abuse Amendments of 1984 and under the Virginia Medical Malpractice Act.

An appropriate order shall issue.

Justice Kennedy

delivered the opinion of the Court.

We address in this ease the application of the Americans with Disabilities Act of 1990 (ADA), 104 Stat. 327,42 U. S. C. § 12101 et seq., to persons infected with the human immunodeficiency virus (HIV). We granted certiorari to review, first, whether HIV infection is a disability under the ADA when the infection has not yet progressed to the so-called symptomatic phase; and, second, whether the Court of Appeals, in affirming a grant of summary judgment, cited sufficient material in the record to determine, as a matter of law, that respondent’s infection with HIV posed no direct threat to the health and safety of her treating dentist. 522 U. S. 991 (1997).

I

Respondent Sidney Abbott (hereinafter respondent) has been infected with HIV since 1986. When the incidents we recite occurred, her infection had not manifested its most serious symptoms. On September 16,1994, she went to the office of petitioner Randon Bragdon in Bangor, Maine, for a dental appointment. She disclosed her HIV infection on the *629patient registration form. Petitioner completed a dental examination, discovered a cavity, and informed respondent of his policy against filling cavities of HIV-infected patients. He offered to perform the work at a hospital with no added fee for his services, though respondent would be responsible for the cost of using the hospital’s facilities. Respondent declined.

Respondent sued petitioner under state law and § 302 of the ADA, 104 Stat. 355,42 U. S. C. § 12182, alleging discrimination on the basis of her disability. The state-law claims are not before us. Section 302 of the ADA provides:

“No individual shall be discriminated against on the basis of disability in the full and equal enjoyment of the goods, services, facilities, privileges, advantages, or accommodations of any place of public accommodation by any person who ... operates a place of public accommodation.” § 12182(a).

The term “public accommodation” is defined to include the “professional office of a health care provider.” § 12181(7)(F).

A later subsection qualifies the mandate not to discriminate. It provides:

“Nothing in this subehapter shall require an entity to permit an individual to participate in or benefit from the goods, services, facilities, privileges, advantages and accommodations of such entity where such individual poses a direct threat to the health or safety of others.” § 12182(b)(3).

The United States and the Maine Human Rights Commission intervened as plaintiffs. After discovery, the parties filed cross-motions for summary judgment. The District Court ruled in favor of the plaintiffs, holding that respondent’s HIV infection satisfied the ADA’s definition of disability. 912 F. Supp. 580, 585-587 (Me. 1995). The court held further that petitioner raised no genuine issue of material fact as to whether respondent’s HIV infection would have *630posed a direct threat to the health or safety of others during the course of a dental treatment. Id., at 587-591. The court relied on affidavits submitted by Dr. Donald Wayne Marianos, Director of the Division of Oral Health of the Centers for Disease Control and Prevention (CDC). The Marianos affidavits asserted it is safe for dentists to treat patients infected with HIV in dental offices if the dentist follows the so-called universal precautions described in the Recommended Infection-Control Practices for Dentistry issued by CDC in 1993 (1993 CDC Dentistry Guidelines). 912 F. Supp., at 589.

The Court of Appeals affirmed. It held respondent’s HIV infection was a disability under the ADA, even though her infection had not yet progressed to the symptomatic stage. 107 F. 3d 934, 939-943 (CA1 1997). The Court of Appeals also agreed that treating the respondent in petitioner’s office would not have posed a direct threat to the health and safety of others. Id, at 943-948. Unlike the District Court, however, the Court of Appeals declined to rely on the Marianos affidavits. Id., at 946, n. 7. Instead the court relied on the 1993 CDC Dentistry Guidelines, as well as the Policy on AIDS, HIV Infection and the Practice of Dentistry, promulgated by the American Dental Association in 1991 (1991 American Dental Association Policy on HIV). 107 F. 3d, at 945-946.

n

We first review the ruling that respondent’s HIV infection constituted a disability under the ADA. The statute defines disability as:

“(A) a physical or mental impairment that substantially limits one or more of the major life activities of such individual;

“(B) a record of such an impairment; or

“(C) being regarded as having such an impairment.” §12102(2).

*631We hold respondent’s HIV infection was a disability under subsection (A) of the definitional section of the statute. In light of this conclusion, we need not consider the applicability of subsections (B) or (C).

Our consideration of subsection (A) of the definition proceeds in three steps. First, we consider whether respondent’s HIV infection was a physical impairment. Second, we identify the life activity upon which respondent relies (reproduction and childbearing) and determine whether it constitutes a major life activity under the ADA. Third, tying the two statutory phrases together, we ask whether the impairment substantially limited the major life activity. In construing the statute, we are informed by interpretations of parallel definitions in previous statutes and the views of various administrative agencies which have faced this interpretive question.

A

The ADA’s definition of disability is drawn almost verbatim from the definition of “handicapped individual” included in the Rehabilitation Act of 1973, 87 Stat. 361, as amended, 29 U. S. C. § 706(8)(B) (1988 ed.), and the definition of “handicap” contained in the Fair Housing Amendments Act of 1988, 102 Stat. 1619, 42 U. S. C. § 3602(h)(1) (1988 ed.). Congress’ repetition of a well-established term carries the implication that Congress intended the term to be construed in accordance with pre-existing regulatory interpretations. See FDIC v. Philadelphia Gear Corp., 476 U. S. 426, 437-438 (1986); Commissioner v. Estate of Noel, 380 U. S. 678, 681-682 (1965); ICC v. Parker, 326 U. S. 60, 65 (1945). In this case, Congress did more than suggest this construction; it adopted a specific statutory provision in the ADA directing as follows:

“Except as otherwise provided in this chapter, nothing in this chapter shall be construed to apply a lesser standard than the standards applied under title V of the *632Rehabilitation Act of 1973 (29 U. S. C. 790 et seq.) or the regulations issued by Federal agencies pursuant to such title.” 42 U.S. C. § 12201(a).

The directive requires us to construe the ADA to grant at least as much protection as provided by the regulations implementing the Rehabilitation Act.

1

The first step in the inquiry under subsection (A) requires us to determine whether respondent’s condition constituted a physical impairment. The Department of Health, Education and Welfare (HEW) issued the first regulations interpreting the Rehabilitation Act in 1977. The regulations are of particular significance because, at the time, HEW was the agency responsible for coordinating the implementation and enforcement of §504 of that statute. Consolidated Rail Corporation v. Darrone, 465 U. S. 624, 634 (1984) (citing Exec. Order No. 11914,3 CFR 117 (1976-1980 Comp.)). Section 504 prohibits discrimination against individuals with disabilities by recipients of federal financial assistance. 29 U. S. C. § 794. The HEW regulations, which appear without change in the current regulations issued by the Department of Health and Human Services, define “physical or mental impairment” to mean:

“(A) any physiological disorder or condition, cosmetic disfigurement, or anatomical loss affecting one or more of the following body systems: neurological; musculo-skeletal; special sense organs; respiratory, including speech organs; cardiovascular; reproductive, digestive, genito-urinary; hemic and lymphatic; skin; and endocrine; or

“(B) any mental or psychological disorder, such as mental retardation, organic brain syndrome, emotional or mental illness, and specific learning disabilities.” 45 CFR §84.3(j)(2)(i) (1997).

*633In issuing these regulations, HEW decided against including a list of disorders constituting physical or mental impairments, out of concern that any specific enumeration might not be comprehensive. 42 Fed. Reg. 22685 (1977), reprinted in 45 CFR pt. 84, App. A, p. 834 (1997). The commentary accompanying the regulations, however, contains a representative list of disorders and conditions constituting physical impairments, including “such diseases and conditions as orthopedic, visual, speech, and hearing impairments, cerebral palsy, epilepsy, muscular dystrophy, multiple sclerosis, cancer, heart disease, diabetes, mental retardation, emotional illness, and ... drug addiction and alcoholism.” Ibid.

In 1980, the President transferred responsibility for the implementation and enforcement of §504 to the Attorney General. See, e. <?., Exec. Order No. 12250, 3 CFR 298 (1981). The regulations issued by the Justice Department, which remain in force to this day, adopted verbatim the HEW definition of physical impairment quoted above. 28 CFR § 41.31(b)(1) (1997). In addition, the representative list of diseases and conditions originally relegated to the commentary accompanying the HEW regulations were incorporated into the text of the regulations. Ibid.

HIV infection is not included in the list of specific disorders constituting physical impairments, in part because HIV was not identified as the cause of AIDS until 1983. See Barré-Sinoussi et al., Isolation of a T-Lymphotropie Retrovirus from a Patient at Risk for Acquired Immune Deficiency Syndrome (AIDS), 220 Science 868 (1983); Gallo et al., Frequent Detection and Isolation of Cytopathie Retroviruses (HTLV-III) from Patients with AIDS and at Risk for AIDS, 224 Science 500 (1984); Levy et al., Isolation of Lymphoeyto-pathic Retroviruses from San Francisco Patients with AIDS, 225 Science 840 (1984). HIV infection does fall well within the general definition set forth by the regulations, however.

The disease follows a predictable and, as of today, an unalterable course. Once a person is infected with HIV, the *634virus invades different cells in the blood and in body tissues. Certain white blood cells, known as helper T-lymphoeytes or CD4+ cells, are particularly vulnerable to HIV. The virus attaches to the CD4 receptor site of the target cell and fuses its membrane to the cell’s membrane. HIV is a retrovirus, which means it uses an enzyme to convert its own genetic material into a form indistinguishable from the genetic material of the target cell. The virus’ genetic material migrates to the cell’s nucleus and becomes integrated with the cell’s chromosomes. Once integrated, the virus can use the cell’s own genetic machinery to replicate itself. Additional copies of the virus are released into the body and infect other cells in turn. Young, The Replication Cycle of HIV-1, in The AIDS Knowledge Base, pp. 3.1-2 to 3.1-7 (P. Cohen, M. Sande, & P. Volberding eds., 2d ed. 1994) (hereinafter AIDS Knowledge Base); Polks & Hart, The Life Cycle of Human Immunodeficiency Virus Type 1, in AIDS: Etiology, Diagnosis, Treatment and Prevention 29-39 (V. DeVita et al. eds., 4th ed. 1997) (hereinafter AIDS: Etiology); Greene, Molecular Insights into HIV-1 Infection, in The Medical Management of AIDS 18-24 (M. Sande & P. Volberding eds., Sth ed. 1997) (hereinafter Medical Management of AIDS). Although the body does produce antibodies to combat HIV infection, the antibodies are not effective in eliminating the virus. Pantaleo et al., Immunopathogenesis of Human Immunodeficiency Virus Infection, in AIDS: Etiology 79; Gardner, HIV Vaccine Development, in AIDS Knowledge Base 3.6-5; Haynes, Immune Responses to Human Immunodeficiency Virus Infection, in AIDS: Etiology 91.

The virus eventually kills the infected host cell. CD4+ eells play a critical role in coordinating the body’s immune response system, and the decline in their number causes corresponding deterioration of the body’s ability to fight infections from many sources. Tracking the infected individual’s CD4+ cell count is one of the most accurate measures of the course of the disease. Greene, Medical Management of *635AIDS 19, 24. Osmond, Classification and Staging of HIV Disease, in AIDS Knowledge Base 1.1-8; Saag, Clinical Spectrum of Human Immunodeficiency Virus Diseases, in AIDS: Etiology 204.

The initial stage of HIV infection is known as acute or primary HIV infection. In a typical case, this stage lasts three months. The virus concentrates in the blood. The assault on the immune system is immediate. The victim suffers from a sudden and serious decline in the number of white blood cells. There is no latency period. Mononucleosis-like symptoms often emerge between six days and six weeks after infection, at times accompanied by fever, headache, enlargement of the lymph nodes (lymphadenopa-thy), muscle pain (myalgia), rash, lethargy, gastrointestinal disorders, and neurological disorders. Usually these symptoms abate within 14 to 21 days. HIV antibodies appear in the bloodstream within 3 weeks; circulating HIV can be detected within 10 weeks. Carr & Cooper, Primary HIV Infection, in Medical Management of AIDS 89-91; Cohen & Volberding, Clinical Spectrum of HIV Disease, in AIDS Knowledge Base 4.1-7; Crowe & McGrath, Acute HIV Infection, in AIDS Knowledge Base 4.2-1 to 4.2-4; Saag, AIDS: Etiology 204-205.

After the symptoms associated with the initial stage subside, the disease enters what is referred to sometimes as its asymptomatic phase. The term is a misnomer, in some respects, for clinical features persist throughout, including lymphadenopathy, dermatological disorders, oral lesions, and bacterial infections. Although it varies with each individual, in most instances this stage lasts from 7 to 11 years. The virus now tends to concentrate in the lymph nodes, though low levels of the virus continue to appear in the blood. Cohen & Volberding, AIDS Knowledge Base 4.1-4, 4.1-8; Saag, AIDS: Etiology 205-206; Staprans & Eeinberg, Natural History and Immunopathogenesis of HIV-1 Disease, in Medical Management of AIDS 29, 38. It was once *636thought the virus became inactive during this period, but it is now known that the relative lack of symptoms is attributable to the virus’ migration from the circulatory system into the lymph nodes. Cohen & Yolberding, AIDS Knowledge Base 4.1-4. The migration reduces the viral presence in other parts of the body, with a corresponding diminution in physical manifestations of the disease. The virus, however, thrives in the lymph nodes, which, as a vital point of the body’s immune response system, represents an ideal environment for the infection of other CD4+ cells. Staprans & Feinberg, Medical Management of AIDS S3-34. Studies have shown that viral production continues at a high rate. Cohen & Yolberding, AIDS Knowledge Base 4.1-4; Stap-rans & Feinberg, Medical Management of AIDS 38. CD4+ cells continue to decline an average of 5% to 10% (40 to 80 eells/mm3) per year throughout this phase. Saag, AIDS: Etiology 207.

A person is regarded as having AIDS when his or her CD4+ count drops below 200 cells/mm3 of blood or when CD4+ cells comprise less than 14% of his or her total lymphocytes. U. S. Dept, of Health and Human Services, Public Health Service, CDC, 1993 Revised Classification System for HIV Infection and Expanded Surveillance Case Definition for AIDS Among Adolescents and Adults, 41 Morbidity and Mortality Weekly Rep., No. RR-17 (Dec. 18, 1992); Osmond, AIDS Knowledge Base 1.1-2; Saag, AIDS: Etiology 207; Ward, Petersen, & Jaffe, Current Trends in the Epidemiology of HIV/AIDS, in Medical Management of AIDS 3. During this stage, the clinical conditions most often associated with HIV, such as pnemnocystis caminii pneumonia, Kaposi’s sarcoma, and non-Hodgkins lymphoma, tend to appear. In addition, the general systemic disorders present during all stages of the disease, such as fever, weight loss, fatigue, lesions, nausea, and diarrhea, tend to worsen. In most cases, once the patient’s CD4+ count drops below 10 *637cells/mm3, death soon follows. Cohen & Yolberding, AIDS Knowledge Base 4.1-9; Saag, AIDS: Etiology 207-209.

In light of the immediacy with which the virus begins to damage the infected person’s white blood cells and the severity of the disease, we hold it is an impairment from the moment of infection. As noted earlier, infection with HIV causes immediate abnormalities in a person’s blood, and the infected person’s white cell count continues to drop throughout the course of the disease, even when the attack is concentrated in the lymph nodes. In light of these facts, HIV infection must be regarded as a physiological disorder with a constant and detrimental effect on the infected person’s hemic and lymphatic systems from the moment of infection. HIV infection satisfies the statutory and regulatory definition of a physical impairment during every stage of the disease.

2

The statute is not operative, and the definition not satisfied, unless the impairment affects a major life activity. Respondent’s claim throughout this case has been that the HIV infection placed a substantial limitation on her ability to reproduce and to bear children. App. 14; 912 F. Supp., at 586; 107 F. 3d, at 939. Given the pervasive, and invariably fatal, course of the disease, its effect on major life activities of many sorts might have been relevant to our inquiry. Respondent and a number of amici make arguments about HIV’s profound impact on almost every phase of the infected person’s life. See Brief for Respondent Abbott 24-27; Brief for American Medical Association as Amicus Curiae 20; Brief for Infectious Diseases Society of America et al. as Amici Curiae 7-11. In light of these submissions, it may seem legalistic to circumscribe our discussion to the activity of reproduction. We have little doubt that had different parties brought the suit they would have maintained that an HIV infection imposes substantial limitations on other major life activities.

*638From the outset, however, the case has been treated as one in which reproduction was the major life activity limited by the impairment. It is our practice to decide cases on the grounds raised and considered in the Court of Appeals and included in the question on which we granted certiorari. See, e. g., Blessing v. Freestone, 520 U. S. 329, 340, n. 3 (1997) (citing this Court’s Rule 14.1(a)); Capitol Square Review and Advisory Bd. v. Pinette, 515 U. S. 753, 760 (1995). We ask, then, whether reproduction is a major life activity.

We have little difficulty concluding that it is. As the Court of Appeals held, “[t]he plain meaning of the word 'major’ denotes comparative importance” and “suggests] that the touchstone for determining an activity’s inclusion under the statutory rubric is its significance.” 107 F. 3d, at 939, 940. Reproduction falls well within the phrase "major life activity.” Reproduction and the sexual dynamics surrounding it are central to the life process itself.

While petitioner concedes the importance of reproduction, he claims that Congress intended the ADA only to cover those aspects of a person’s life which have a public, economic, or daily character. Brief for Petitioner 14, 28, 30, 31; see also id., at 36-37 (citing Krauel v. Iowa Methodist Medical Center, 95 F. 3d 674, 677 (CA8 1996)). The argument founders on the statutory language. Nothing in the definition suggests that activities without a public, economic, or daily dimension may somehow be regarded as so unimportant or insignificant as to fall outside the meaning of the word “major.” The breadth of the term confounds the attempt to limit its construction in this maimer.

As we have noted, the ADA must be construed to be consistent with regulations issued to implement the Rehabilitation Act. See 42 U. S. C. § 12201(a). Rather than enunciating a general principle for determining what is and is not a major life activity, the Rehabilitation Act regulations instead provide a representative list, defining the term to include "functions such as caring for one’s self, performing manual *639tasks, walking, seeing, hearing, speaking, breathing, learning, and working.” 45 CFR §84.3(j)(2)(ii) (1997); 28 CFR § 41.31(b)(2) (1997). As the use of the term "such as” confirms, the list is illustrative, not exhaustive.

These regulations are contrary to petitioner’s attempt to limit the meaning of the term “major” to public activities. The inclusion of activities such as caring for one’s self and performing manual tasks belies the suggestion that a task must have a public or economic character in order to be a major life activity for purposes of the ADA. On the contrary, the Rehabilitation Act regulations support the inclusion of reproduction as a major life activity, since reproduction could not be regarded as any less important than working and learning. Petitioner advances no credible basis for confining major life activities to those with a public, economic, or daily aspect. In the absence of any reason to reach a contrary conclusion, we agree with the Court of Appeals’ determination that reproduction is a major life activity for the purposes of the ADA.

3

The final element of the disability definition in subsection (A) is whether respondent’s physical impairment was a substantial limit on the major life activity she asserts. The Rehabilitation Act regulations provide no additional guidance. 45 CFR pt. 84, App. A, p. 334 (1997).

Our evaluation of the medical evidence leads us to conclude that respondent’s infection substantially limited her ability to reproduce in two independent ways. First, a woman infected with HIY who tries to conceive a child imposes on the man a significant risk of becoming infected. The cumulative results of 13 studies collected in a 1994 textbook on AIDS indicates that 20% of male partners of women with HIV became HIV-positive themselves, with a majority of the studies finding a statistically significant risk of infection. Osmond & Padian, Sexual Transmission of HIV, in AIDS *640Knowledge Base 1.9-8, and tbl. 2; see also Haverkos & Battjes, Female-to-Male Transmission of HIV, 268 JAMA 1855, 1856, tbl. (1992) (cumulative results of 16 studies indicated 25% risk of female-to-male transmission). (Studies report a similar, if not more severe, risk of male-to-female transmission. See, e.g., Osmond & Padian, AIDS Knowledge Base 1.9-8, tbl. 1,1.9-6 to 1.9-7.)

Second, an infected woman risks infecting her child during gestation and childbirth, i. e., perinatal transmission. Petitioner concedes that women infected with HIV face about a 25% risk of transmitting the virus to their children. 107 F. 3d, at 942; 912 F. Supp., at 587, n. 6. Published reports available in 1994 confirm the accuracy of this statistic. Report of a Consensus Workshop, Maternal Factors Involved in Mother-to-Child Transmission of HIV-1, 5 J. Acquired Immune Deficiency Syndromes 1019, 1020 (1992) (collecting 13 studies placing risk between 14% and 40%, with most studies falling within the 25% to 30% range); Connor et al., Reduction of Maternal-Infant Transmission of Human Immunodeficiency Virus Type 1 with Zidovudine Treatment, 331 New England J. Med. 1173,1176 (1994) (placing risk at 25.5%); see also Staprans & Feinberg, Medical Management of AIDS 32 (studies report 13% to 45% risk of infection, with average of approximately 25%).

Petitioner points to evidence in the record suggesting that antiretroviral therapy can lower the risk of perinatal transmission to about 8%. App. 53; see also Connor, supra, at 1176 (8.3%); Sperling et al., Maternal Viral Load, Zidovudine Treatment, and the Risk of Transmission of Human Immunodeficiency Virus Type 1 from Mother to Infant, 335 New England J. Med. 1621,1622 (1996) (7.6%). The United States questions the relevance of the 8% figure, pointing to regulatory language requiring the substantiality of a limitation to be assessed without regard to available mitigating measures. Brief for United States as Amicus Curiae 18, n. 10 (citing 28 CFR pt. 36, App. B, p. 611 (1997); 29 CFR pt. *6411630, App., p. 351 (1997)). We need not resolve this dispute in order to decide this case, however. It cannot he said as a matter of law that an 8% risk of transmitting a dread and fatal disease to one’s child does not represent a substantial limitation on reproduction.

The Act addresses substantial limitations on major life activities, not utter inabilities. Conception and childbirth are not impossible for an HIV victim but, without doubt, are dangerous to the public health. This meets the definition of a substantial limitation. The decision to reproduce carries economic and legal consequences as well. There are added costs for antiretroviral therapy, supplemental insurance, and long-term health care for the child who must be examined and, tragic to think, treated for the infection. The laws of some States, moreover, forbid persons infected with HIV to have sex with others, regardless of consent. Iowa Code §§139.1, 139.31 (1997); Md. Health Code Ann. §18-601.1(a) (1994); Mont. Code Ann. §§50-18-101, 50-18-112 (1997); Utah Code Ann. §26-6-3.5(3) (Supp. 1997); id., §26-6-5 (1995); Wash. Rev. Code § 9A.36.011(l)(b) (Supp. 1998); see also N. D. Cent. Code §12.1-20-17 (1997).

In the end, the disability definition does not turn on personal choice. When significant limitations result from the impairment, the definition is met even if the difficulties are not insurmountable. For the statistical and other reasons we have cited, of course, the limitations on reproduction may be insurmountable here. Testimony from the respondent that her HIV infection controlled her decision not to have a child is unchallenged. App. 14; 912 F. Supp., at 587; 107 F. 3d, at 942. In the context of reviewing summary judgment, we must take it to be true. Fed. Rule Civ. Proc. 56(e). We agree with the District Court and the Court of Appeals that no triable issue of fact impedes a ruling on the question of statutory coverage. Respondent’s HIV infection is a physical impairment which substantially limits a major life activity, as the ADA defines it. In view of our holding, we *642need not address the second question presented, i. e., whether HIV infection is a per se disability under the ADA.

B

Our holding is confirmed by a consistent course of agency interpretation before and after enactment of the ADA. Every agency to consider the issue under the Rehabilitation Act found statutory coverage for persons with asymptomatic HIV. Responsibility for ádministering the Rehabilitation Act was not delegated to a single agency, but we need not pause to inquire whether this causes us to withhold deference to agency interpretations under Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837, 844 (1984). It is enough to observe that the well-reasoned views of the agencies implementing a statute “constitute a body of experience and informed judgment to which courts and litigants may properly resort for guidance.” Skidmore v. Swift & Co., 323 U. S. 134, 139-140 (1944).

One comprehensive and significant administrative precedent is a 1988 opinion issued by the Office of Legal Counsel of the Department of Justice (OLC) concluding that the Rehabilitation Act “protects symptomatic and asymptomatic HIV-infected individuals against discrimination in any covered program.” Application of Section 504 of the Rehabilitation Act to HIV-Infected Individuals, 12 Op. Off. Legal Counsel 264, 264-265 (Sept. 27, 1988) (preliminary print) (footnote omitted). Relying on a letter from Surgeon General C. Everett Koop stating that, “from a purely scientific perspective, persons with HIV are elearly impaired” even during the asymptomatic phase, OLC determined asymptomatic HIV was a physical impairment under the Rehabilitation Act because it constituted a “physiological disorder or condition affecting the hemic and lymphatic systems.” Id., at 271 (internal quotation marks omitted). OLC determined further that asymptomatic HIV imposed a substantial limit on the major life activity of reproduction. The opinion said:

*643“Based on the medical knowledge available to us, we believe that it is reasonable to conclude that the life activity of procreation ... is substantially limited for an asymptomatic HIV-infected individual. In light of the significant risk that the AIDS virus may be transmitted to a baby during pregnancy, HIV-infected individuals cannot, whether they are male or female, engage in the act of procreation with the normal expectation of bringing forth a healthy child.” Id., at 273.

In addition, OLC indicated that “[t]he life activity of engaging in sexual relations is threatened and probably substantially limited by the contagiousness of the virus.” Id., at 274. Either consideration was sufficient to render asymptomatic HIV infection a handicap for purposes of the Rehabilitation Act. In the course of its opinion, OLC considered, and rejected, the contention that the limitation could be discounted as a voluntary response to the infection. The limitation, it reasoned, was the infection’s manifest physical effect. Id., at 274, and n. 13. Without exception, the other agencies to address the problem before enactment of the ADA reached the same result. Federal Contract Compliance Manual App. 6D, 8 FEP Manual 405:352 (Dec. 23,1988); In re Ritter, No. 03890089, 1989 WL 609697, *10 (EEOC, Dec. 8, 1989); see also Comptroller General’s Task Force on AIDS in the Workplace, Coping with AIDS in the GAO Workplace: Task Force Report 29 (Dec. 1987); Report of the Presidential Commission on the Human Immunodeficiency Virus Epidemic 113-114,122-123 (June 1988). Agencies have adhered to this conclusion since the enactment of the ADA as well. See 5 CFR § 1636.103 (1997); 7 CFR § 15e.l03 (1998); 22 CFR §1701.103 (1997); 24 CFR §9.103 (1997); 34 CFR §1200.103 (1997); 45 CFR §§2301.103,2490.103 (1997); In re Westchester County Medical Center, [1991-1994 Transfer Binder] CCH Employment Practices Guide ¶ 5340, pp. 6110-6112 (Apr. 20, 1992), aff’d, id., ¶5362, pp. 6249-6250 (Dept, of Health & Human Servs. Departmental Appeals Bd., Sept. 25, 1992); *644In re Rosebud Sioux Tribe, No. 93-504-1, 1994 WL 603015 (Dept. of Health & Human Servs. Departmental Appeals Bd., July 14, 1994); In re Martin, No. 01954089, 1997 WL 151524, *4 (EEOC, Mar. 27, 1997).

Every court which addressed the issue before the ADA was enacted in July 1990, moreover, concluded that asymptomatic HIV infection satisfied the Rehabilitation Act’s definition of a handicap. See Doe v. Garrett, 903 F. 2d 1455, 1457 (CA11 1990), cert. denied, 499 U. S. 904 (1991); Ray v. School Dist. of DeSoto County, 666 F. Supp. 1524, 1536 (MD Fla. 1987); Thomas v. Atascadero Unified School Dist., 662 F. Supp. 376, 381 (CD Cal. 1987); District 27 Community School Bd. v. Board of Ed. of New York, 130 Misc. 2d 398, 413-415, 502 N. Y. S. 2d 325, 335-337 (Sup. Ct., Queens Cty. 1986); cf. Baxter v. Belleville, 720 F. Supp. 720, 729 (SD Ill. 1989) (Fair Housing Amendments Act); Cain v. Hyatt, 734 F. Supp. 671, 679 (ED Pa. 1990) (Pennsylvania Human. Relations Act). (For cases finding infection with HIV to be a handicap without distinguishing between symptomatic and asymptomatic HIV, see Martinez ex rel. Martinez v. School Bd. of Hillsborough Cty., 861 F. 2d 1502, 1506 (CA11 1988); Chalk v. United States Dist. Ct., 840 F. 2d 701, 706 (CA9 1988); Doe v. Dolton Elementary School Dist. No. 148, 694 F. Supp. 440, 444-445 (ND Ill. 1988); Robertson v. Granite City Community Unit School Dist. No. 9, 684 F. Supp. 1002, 1006-1007 (SD Ill. 1988); Local 1812, AFGE v. United States Dept. of State, 662 F. Supp. 50, 54 (DC 1987); cf. Association of Relatives and Friends of AIDS Patients v. Regulations and Permits Admin., 740 F. Supp. 95, 103 (PR 1990) (Fair Housing Amendments Act).) We are aware of no instance prior to the enactment of the ADA in which a court or agency ruled that HIV infection was not a handicap under the Rehabilitation Act.

Had Congress done nothing more than copy the Rehabilitation Act definition into the ADA, its action would indicate *645the new statute should be construed in light of this unwavering line of administrative and judicial interpretation. All indications are that Congress was well aware of the position taken by OLC when enacting the ADA and intended to give that position its active endorsement. H. R. Rep. No. 101-485, pt. 2, p. 52 (1990) (endorsing the analysis and conclusion of the OLC opinion); id., pt. 3, at 28, n. 18 (same); S. Rep. No. 101-116, pp. 21,22 (1989) (same). As noted earlier, Congress also incorporated the same definition into the Fair Housing Amendments Act of 1988. See 42 U. S. C. § 3602(h)(1). We find it significant that the implementing regulations issued by the Department of Housing and Urban Development (HUD) construed the definition to include infection with HIV. 54 Fed. Reg. 3232, 3245 (1989) (codified at 24 CFR §100.201 (1997)); see also In re Williams, 2A P-H Fair Housing-Fair Lending ¶ 25,007, pp. 25,111-25,113 (HUD Off. Admin. Law Judges, Mar. 22, 1991) (adhering to this interpretation); In re Elroy B. and Dorothy Burns Trust, 2A P-H Fair Housing-Fair Lending ¶ 25,073, p. 25,678 (HUD Off. Admin. Law Judges, June 17, 1994) (same). Again the legislative record indicates that Congress intended to ratify HUD’s interpretation when it reiterated the same definition in the ADA. H. R. Rep. No. 101-485, pt. 2, at 50; id., pt. 3, at 27; id., pt. 4, at 36; S. Rep. No. 101-116, at 21.

We find the uniformity of the administrative and judicial precedent construing the definition significant. When administrative and judicial interpretations have settled the meaning of an existing statutory provision, repetition of the same language in a new statute indicates, as a general matter, the intent to incorporate its administrative and judicial interpretations as well. See, e. g., Lorillard v. Pons, 434 U. S. 575, 580-581 (1978). The uniform body of administrative and judicial precedent confirms the conclusion we reach today as the most faithful way to effect the congressional design.

*646c

Our conclusion is further reinforced by the administrative guidance issued by the Justice Department to implement the public accommodation provisions of Title III of the ADA. As the agency directed by Congress to issue implementing regulations, see 42 U. S. C. § 12186(b), to render technical assistance explaining the responsibilities of covered individuals and institutions, § 12206(c), and to enforce Title III in court, § 12188(b), the Department’s views are entitled to deference. See Chevron, 467 U. S., at 844.

The Justice Department’s interpretation of the definition of disability is consistent with our analysis. The regulations acknowledge that Congress intended the ADA’s definition of disability to be given the same construction as the definition of handicap in the Rehabilitation Act. 28 CFR § 86.103(a) (1997); id., pt. 86, App. B, pp. 608, 609. The regulatory definition developed by HEW to implement the Rehabilitation Act is incorporated verbatim in the ADA regulations. §36.104. The Justice Department went further, however. It added “HIV infection (symptomatic and asymptomatic)” to the list of disorders constituting a physical impairment. § 36.104(l)(iii). The technical assistance the Department has issued pursuant to 42 U. S. C. § 12206 similarly concludes that persons with asymptomatic HIV infection fall within the ADA’s definition of disability. See, e. g., U. S. Dept, of Justice, Civil Rights Division, The Americans with Disabilities Act: Title III Technical Assistance Manual 9 (Nov. 1993); Response to Congressman Sonny Callahan, 5 Nat. Disability L. Rep. (LRP) ¶ 360, p. 1167 (Feb. 9,1994); Response to A. Laurence Field, 5 Nat. Disaoility L. Rep. (LRP) ¶ 21, p. 80 (Sept. 10,1993). Any other conclusion, the Department reasoned, would contradict Congress’ affirmative ratification of the administrative interpretations given previous versions of the same definition. 28 CFR pt. 36, App. B, pp. 609, 610 (1997) (citing the OLC opinion and HUD regulations); 56 Fed. Reg. 7455, 7456 (1991) (same) (notice of proposed rulemaking).

*647We also draw guidance from the views of the agencies authorized to administer other sections of the ADA. See 42 U. S. C. § 12116 (authorizing EEOC to issue regulations implementing Title I); § 12134(a) (authorizing the Attorney General to issue regulations implementing the public services provisions of Title II, subtitle A); §§12149, 12164, 12186 (authorizing the Secretary of Transportation to issue regulations implementing the transportation-related provisions of Titles II and III); § 12206(c) (authorizing the same agencies to offer technical assistance for the provisions they administer). These agencies, too, concluded that HIV infection is a physical impairment under the ADA. 28 CPR §35.104(l)(iii) (1997); 49 CFR §§37.3, 38.3 (1997); 56 Fed. Reg. 13858 (1991); U. S. Dept, of Justice, Civil Rights Division, The Americans with Disabilities Act: Title II Technical Assistance Manual 4 (Nov. 1993); EEOC, A Technical Assistance Manual on the Employment Provisions (Title I) of the Americans with Disabilities Act II-3 (Jan. 1992) (hereinafter EEOC Technical Assistance Manual); EEOC Interpretive Manual § 902.2(d), pp. 902-13 to 902-14 (reissued Mar. 14, 1995) (hereinafter EEOC Interpretive Manual), reprinted in 2 BNA EEOC Compliance Manual 902:0013 (1998). Most categorical of all is EEOC’s conclusion that “an individual who has HIV infection (including asymptomatic HIV infection) is an individual with a disability.” EEOC Interpretive Manual § 902.4(c)(1), p. 902-21; accord, id., § 902.2(d), p. 902-14, n. 18. In the EEOC’s view, “impairments . . . such as HIV infection, are inherently substantially limiting.” 29 CFR pt. 1630, App., p. 350 (1997); EEOC Technical Assistance Manual II — 4; EEOC Interpretive Manual § 902.4(c)(1), p. 902-21.

The regulatory authorities we cite are consistent with our holding that HIV infection, even in the so-called asymptomatic phase, is an impairment which substantially limits the major life activity of reproduction.

*648r — i hH í — Í

The petition for certiorari presented three other questions for review. The questions stated:

“3. When deciding under title III of the ADA whether a private health care provider must perform invasive procedures on an infectious patient in his office, should courts defer to the health care provider’s professional judgment, as long as it is reasonable in light of then-current medical knowledge?

"4. What is the proper standard of judicial review under title III of the ADA of a private health care provider’s judgment that the performance of certain invasive procedures in his office would pose a direct threat to the health or safety of others?

“5. Did petitioner, Randon Bragdon, D. M. D., raise a genuine issue of fact for trial as to whether he was warranted in his judgment that the performance of certain invasive procedures on a patient in his office would have posed a direct threat to the health or safety of others?” Pet. for Cert. i.

Of these, we granted certiorari only on question three. The question is phrased in an awkward way, for it conflates two separate inquiries. In asking whether it is appropriate to defer to petitioner’s judgment, it assumes that petitioner’s assessment of the objective facts was reasonable. The central premise of the question and the assumption on which it is based merit separate consideration.

Again, we begin with the statute. Notwithstanding the protection given respondent by the ADA’s definition of disability, petitioner could have refused to treat her if her infectious condition “pose[d] a direct threat to the health or safety of others.” 42 U. S. C. § 12182(b)(3). The ADA defines a direct threat to be “a significant risk to the health or safety of others that cannot be eliminated by a modification of policies, practices, or procedures or by the provision of auxiliary aids *649or services.” Ibid. Parallel provisions appear in the employment provisions of Title I §§ 12111(8), 12113(b).

The ADA’s direct threat provision stems from the recognition in School Bd. of Nassau Cty. v. Arline, 480 U. S. 273, 287 (1987), of the importance of prohibiting discrimination against individuals with disabilities while protecting others from significant health and safety risks, resulting, for instance, from a contagious disease. In Arline, the Court reconciled these objectives by construing the Rehabilitation Act not to require the hiring of a person who posed “a significant risk of communicating an infectious disease to others.” Id., at 287, n. 16. Congress amended the Rehabilitation Act and the Fair Housing Act to incorporate the language. See 29 U. S. C. §706(8)(D) (excluding individuals who “would constitute a direct threat to the health or safety of other individuals”); 42 U. S. C. § 3604(f)(9) (same). It later relied on the. same language in enacting the ADA. See 28 CFR pt. 36, App. B, p. 626 (1997) (ADA’s direct threat provision codifies Arline). Because few, if any, activities in life are risk free, Arline and the ADA do not ask whether a risk exists, but whether it is significant. Arline, supra, at 287, and n. 16; 42 Ü.S.C. § 12182(b)(3).

The existence, or nonexistence, of a significant risk must be determined from the standpoint of the person who refuses the treatment or accommodation, and the risk assessment must be based on medical or other objective evidence. Arline, supra, at 288; 28 CFR § 36.208(c) (1997); id., pt. 36, App. B, p. 626. As a health care professional, petitioner had the duty to assess the risk of infection based on the objective, scientific information available to him and others in his profession. His belief that a significant risk existed, even if maintained in good faith, would not relieve him from liability. To use the words of the question presented, petitioner receives no special deference simply because he is a health care professional. It is true that Arline reserved “the question whether courts should also defer to the reasonable medical *650judgments of private physicians on which an employer has relied.” 480 U. S., at 288, n. 18. At most, this statement reserved the possibility that employers could consult with individual physicians as objective third-party experts. It did not suggest that an individual physician’s state of mind could excuse discrimination without regard to the objective reasonableness of his actions.

Our conclusion that courts should assess the objective reasonableness of the views of health care professionals without deferring to their individual judgments does not answer the implicit assumption in the question presented, whether petitioner’s actions were reasonable in light of the available medical evidence. In assessing the reasonableness of petitioner’s actions, the views of public health authorities, such as the U. S. Public Health Service, CDC, and the National Institutes of Health, are of special weight and authority. Arline, supra, at 288; 28 CFR pt. 36, App. B, p. 626 (1997). The views of these organizations are not conclusive, however. A health care professional who disagrees with the prevailing medical consensus may refute it by citing a credible scientific basis for deviating from the accepted norm. See W. Keeton, D. Dobbs, R. Keeton, & D. Owen, Prosser and Keeton on Law of Torts § 32, p. 187 (5th ed. 1984).

We have reviewed so much of the record as necessary to illustrate the application of the rule to the facts of this case. For the most part, the Court of Appeals followed the proper standard in evaluating petitioner’s position and conducted a thorough review of the evidence. Its rejection of the District Court’s reliance on the Marianos affidavits was a correct application of the principle that petitioner’s actions must be evaluated in light of the available, objective evidence. The record did not show that CDC had published the conclusion set out in the affidavits at the time petitioner refused to treat respondent. 107 F. 3d, at 946, n. 7.

A further illustration of a correct application of the objective standard is the Court of Appeals’ refusal to give weight *651to petitioner’s offer to treat respondent in a hospital. Id., at 943, n. 4. Petitioner testified that he believed hospitals had safety measures, such as air filtration, ultraviolet lights, and respirators, which would reduce the risk of HIV transmission. App. 151. Petitioner made no showing, however, that any area hospital had these safeguards or even that he had hospital privileges. Id., at 31. His expert also admitted the lack of any scientific basis for the conclusion that these measures would lower the risk of transmission. Id., at 209. Petitioner failed to present any objective, medical evidence showing that treating respondent in a hospital would be safer or more efficient in preventing HIV transmission than treatment in a well-equipped dental office.

We are concerned, however, that the Court of Appeals might have placed mistaken reliance upon two other sources. In ruling no triable issue of fact existed on this point, the Court of Appeals relied on the 1993 CDC Dentistry Guidelines and the 1991 American Dental Association Policy on HIV. 107 F. 3d, at 945-946. This evidence is not definitive. As noted earlier, the CDC Guidelines recommended certain universal precautions which, in CDC’s view, “should reduce the risk of disease transmission in the dental environment.” U. S. Dept, of Health and Human Services, Public Health Service, CDC, Recommended Infection-Control Practices for Dentistry, 41 Morbidity and Mortality Weekly Rep. No. RR-8, p. 1 (May 28,1993). The Court of Appeals determined that, “[w]hile the guidelines do not state explicitly that no further risk-reduction measures are desirable or that routine dental care for HIV-positive individuals is safe, those two conclusions seem to be implicit in the guidelines’ detailed delineation of procedures for office treatment of HIV-positive patients.” 107 F. 3d, at 946. In our view, the Guidelines do not necessarily contain implicit assumptions conclusive of the point to be decided. The Guidelines set out CDG’s recommendation that the universal precautions are the best way *652to combat the risk of HIV transmission. They do not assess the level of risk.

Nor can we be certain, on this record, whether the 1991 American Dental Association Policy on HIV carries the weight the Court of Appeals attributed to it. The Policy does provide some evidence of the medical community’s objective assessment of the risks posed by treating people infected with HIV in dental offices. It indicates:

"Current scientific and epidemiologic evidence indicates that there is little risk of transmission of infectious diseases through dental treatment if recommended infection control procedures are routinely followed. Patients with HIV infection may be safely treated in private dental offices when appropriate infection control procedures are employed. Such infection control procedures provide protection both for patients and dental personnel.” App. 225.

We note, however, that the Association is a professional organization, which, although a respected source of information on the dental profession, is not a public health authority. It is not clear the extent to which the Policy was based on the Association’s assessment of dentists’ ethical and professional duties in addition to its scientific assessment of the risk to which the ADA refers. Efforts to clarify dentists’ ethical obligations and to encourage dentists to treat patients with HIV infection with compassion may be commendable, but the question under the statute is one of statistical likelihood, not professional responsibility. Without more information on the manner in which the American Dental Association formulated this Policy, we are unable to determine the Policy’s value in evaluating whether petitioner’s assessment of the risks was reasonable as a matter of law.

The court considered materials submitted by both parties on the cross-motions for summary judgment. The petitioner was required to establish that there existed a genuine *653issue of material fact. Evidence which was merely colorable or not significantly probative would not have been sufficient. Anderson v. Liberty Lobby, Inc., 477 U. S. 242, 249-250 (1986).

We acknowledge the presence of other evidence in the record before the Court of Appeals which, subject to further arguments and examination, might support affirmance of the trial court’s ruling. For instance, the record contains substantial testimony from numerous health experts indicating that it is safe to treat patients infected with HIV in dental offices. App. 66-68,88-90,264-266,268. We are unable to determine the import of this evidence, however. The record does not disclose whether the expert testimony submitted by respondent turned on evidence available in September 1994. See id., at 69-70 (expert testimony relied in part on materials published after September 1994).

There are reasons to doubt whether petitioner advanced evidence sufficient to raise a triable issue of fact on the significance of the risk. Petitioner relied on two principal points: First, he asserted that the use of high-speed drills and surface cooling with water created a risk of airborne HIV transmission. The study on which petitioner relied was inconclusive, however, determining only that “[fjurther work is required to determine whether such a risk exists.” Johnson & Robinson, Human Immunodeficiency Virus-1 (HIV-1) in the Vapors of Surgical Power Instruments, 33 J. of Medical Virology 47 (1991). Petitioner’s expert witness conceded, moreover, that no evidence suggested the spray could transmit HIV. His opinion on airborne risk was based on the absence of contrary evidence, not on positive data. App. 166. Scientific evidence and expert testimony must have a traceable, analytical basis in objective fact before it may be considered on summary judgment. See General Electric Co. v. Joiner, 522 U.S. 136, 144-145, 146 (1997).

Second, petitioner argues that, as of September 1994, CDC had identified seven dental workers with possible oceupa-*654tional transmission of HIV. See U. S. Dept, of Health and Human Services, Public Health Service, CDC, HIV/AIDS Surveillance Report, vol. 6, no. 1, p. 15, tbl. 11 (Mid-year ed. June 1994). These dental workers were exposed to HIV in the course of their employment, but CDC could not determine whether HIV infection had resulted from this exposure. Id., at 15, n. 3. It is now known that CDC could not ascertain how the seven dental workers contracted the disease because they did not present themselves for HIV testing at an appropriate time after this occupational exposure. Gooch et al., Percutaneous Exposures to HIV-Infeeted Blood Among Dental Workers Enrolled in the CDC Needlestick Study, 126 J. American Dental Assn. 1237, 1239 (1995). It is not clear on this record, however, whether this information was available to petitioner in September 1994. If not, the seven eases might have provided some, albeit not necessarily sufficient, support for petitioner’s position. Standing alone, we doubt it would meet the objective, scientific basis for finding a significant risk to the petitioner.

Our evaluation of the evidence is constrained by the fact that on these and other points we have not had briefs and arguments directed to the entire record. In accepting the case for review, we declined to grant certiorari on question five, which asked whether petitioner raised a genuine issue of fact for trial. Pet. for Cert. i. As a result, the briefs and arguments presented to us did not concentrate on the question of sufficiency in light all of the submissions in the summary judgment proceeding. “When attention has been focused on other issues, or when the court from which a case comes has expressed no views on a controlling question, it may be appropriate to remand the ease rather than deal with the merits of that question in this Court.” Dandridge v. Williams, 397 U. S. 471, 476, n. 6 (1970). This consideration carries particular force where, as here, full briefing directed at the issue would help place a complex factual record in proper perspective. Resolution of the issue will be of im-*655portanee to health care workers not just for the result but also for the precision and comprehensiveness of the reasons given for the decision.

We conclude the proper course is to give the Court of Appeals the opportunity to determine whether our analysis of some of the studies cited by the parties would change its conclusion that petitioner presented neither objective evidence nor a triable issue of fact on the question of risk. In remanding the ease, we do not foreclose the possibility that the Court of Appeals may reach the same conclusion it did earlier. A remand will permit a full exploration of the issue through the adversary process.

The determination of the Court of Appeals that respondent’s HIV infection was a disability under the ADA is affirmed. The judgment is vacated, and the case is remanded for further proceedings consistent with this opinion.

It is so ordered.

Justice Stevens,

with whom Justice Breyer joins, concurring.

The Court’s opinion demonstrates that respondent’s HIV infection easily falls within the statute’s definition of “disability.” Moreover, the Court’s discussion in Part III of the relevant evidence has persuaded me that the judgment of the Court of Appeals should be affirmed. I do not believe petitioner has sustained his burden of adducing evidence sufficient to raise a triable issue of fact on the significance of the risk posed by treating respondent in his office. The Court of Appeals reached that conclusion after a careful and extensive study of the record; its analysis on this question was perfectly consistent with the legal reasoning in Justice Kennedy’s opinion for the Court; and the latter opinion itself explains that petitioner relied on data that were inconclusive and speculative at best, see ante, at 653-654. Cf. General Electric Co. v. Joiner, 522 U. S. 136 (1997).

*656There are not, however, five Justices who agree that the judgment should be affirmed. Nor does it appear that there are five Justices who favor a remand for further proceedings consistent with the views expressed in either Justice Kennedy’s opinion for the Court or the opinion of The Chief Justice. Because I am in agreement with the legal analysis in Justice Kennedy’s opinion, in order to provide a judgment supported by a majority, I join that opinion even though I would prefer an outright affirmance. Cf. Screws v. United States, 325 U. S. 91, 134 (1945) (Rutledge, J., concurring in result).

Justice Ginsburg,

concurring.

Human Immunodeficiency Virus (HIV) infection, as the description set out in the Court’s opinion documents, ante, at 635-637, has been regarded as a disease limiting life itself. See Brief for American Medical Association as Amicus Curiae 20. The disease inevitably pervades life’s choices: education, employment, family and financial undertakings. It affects the need for and, as this case shows, the ability to obtain health care because of the reaction of others to the impairment. No rational legislator, it seems to me apparent, would require nondiscrimination once symptoms become visible but permit discrimination when the disease, though present, is not yet visible. I am therefore satisfied that the statutory and regulatory definitions are well met. HIV infection is “a physical... impairment that substantially limits . . . major life activities,” or is so perceived, 42 U. S. C. §§ 12102(2)(A), (C), including the afflicted individual’s family relations, employment potential, and ability to care for herself, see 45 CFR § 84.3(j)(2)(ii) (1997); 28 CFR § 41.31(b)(2) (1997).

I further agree, in view of the “importance [of the issue] to health care workers,” ante, at 654-655, that it is wise to remand, erring, if at all, on the side of caution. By taking this course, the Court ensures a fully informed determina*657tion whether respondent Abbott’s disease posed “a significant risk to the health or safety of [petitioner Bragdon] that [could not] be eliminated by a modification of policies, practices, or procedures ...42 U. S. C. § 12182(b)(8).

Chief Justice Rehnquist,

with whom Justice Scalia and Justice Thomas join, and with whom Justice O’Con-nor joins as to Part II, concurring in the judgment in part and dissenting in part.

I

Is respondent Abbott (hereinafter respondent) — who has tested positive for the human immunodeficiency virus (HIV) but was asymptomatic at the time she suffered discriminatory treatment — a person with a “disability” as that term is defined in the Americans with Disabilities Act of 1990 (ADA)? The term “disability” is defined in the ADA to include:

“(A) a physical or mental impairment that substantially limits one or more of the major life activities of such individual;

“(B) a record of such an impairment; or

“(C) being regarded as having such an impairment.” 42 U.S.C. §12102(2).

It is important to note that whether respondent has a disability covered by the ADA is an individualized inquiry. The Act could not be clearer on this point: Section 12102(2) states explicitly that the disability determination must be made “with respect to an individual.” Were this not sufficiently clear, the Act goes on to provide that the “major life activities” allegedly limited by an impairment must be those “of such individual.” § 12102(2)(A).

The individualized nature of the inquiry is particularly important in this ease because the District Court disposed of it on summary judgment. Thus all disputed issues of material fact must be resolved against respondent. She contends *658that her asymptomatic HIV status brings her within the first definition of a “disability.”1 She must therefore demonstrate, inter alia, that she was (1) physically or mentally impaired and that such impairment (2) substantially limited (3) one or more of her major life activities.

Petitioner does not dispute that asymptomatic HIV-positive status is a physical impairment. I therefore assume this to be the case, and proceed to the second and third statutory requirements for “disability.”

According to the Court, the next question is “whether reproduction is a major life activity.” Ante, at 638. That, however, is only half of the relevant question. As mentioned above, the ADA's definition of a “disability5' requires that the major life activity at issue be one “of such individual.” § 12102(2)(A). The Court truncates the question, perhaps because there is not a shred of record evidence indicating that, prior to becoming infected with HIV, respondent’s major life activities included reproduction2 (assuming *659for the moment that reproduction is a major life activity at all). At most, the record indicates that after learning of her HIV status, respondent, whatever her previous inclination, conclusively decided that she would not have children. App. 14. There is absolutely no evidence that, absent the HIV, respondent would have had or was even considering having children. Indeed, when asked during her deposition whether her HIV infection had in any way impaired her ability to carry out any of her life functions, respondent answered “No.” Ibid. It is further telling that in the course of her entire brief to this Court, respondent studiously avoids asserting even once that reproduction is a major life activity to her. To the contrary, she argues that the “major life activity” inquiry should not turn on a particularized assessment of the circumstances of this or any other case. Brief for Respondent Abbott 30-31.

But even aside from the facts of this particular ease, the Court is simply wrong in concluding as a general matter that reproduction is a “major life activity.” Unfortunately, the ADA does not define the phrase “major life activities.” But the Act does incorporate by reference a list of such activities contained in regulations issued under the Rehabilitation Act. 42 U.S.C. § 12201(a); 45 CFR § 84.3(j)(2)(ii) (1997). The Court correctly recognizes that this list of major life activities “is illustrative, not exhaustive,” ante, at 639, but then makes no attempt to demonstrate that reproduction is a major life activity in the same sense that “caring for one’s self, performing manual tasks, walking, seeing, hearing, speaking, breathing, learning, and working” are. Ante, at 638-639.

Instead, the Court argues that reproduction is a “major” life activity in that it is “central to the life process itself.” Ante, at 638. In support of this reading, the Court focuses on the fact that “'major’” indicates “'comparative impor*660tance,’” ibid.; see also Webster’s Collegiate Dictionary 702 (10th ed. 1994) (“greater in dignity, rank, importance, or interest”), ignoring the alternative definition of “major” as “greater in quantity, number, or extent,” ibid. It is the latter definition that is most consistent with the ADA’s illustrative list of major life activities.

No one can deny that reproductive decisions are important in a person’s life. But so are decisions as to who to many, where to live, and how to earn one’s living. Fundamental importance of this sort is not the common thread linking the statute’s listed activities. The common thread is rather that the activities are repetitively performed and essential in the day-to-day existence of a normally functioning individual. They are thus quite different from the series of activities leading to the birth of a child.

Both respondent, Brief for Respondent Abbott 20, n, 24, and the Government, Brief for United States as Amicus Curiae 18, argue that reproduction must be a major life activity because regulations issued under the ADA define the term “physical impairment” to include physiological disorders affecting the reproductive system. 28 CFR §36.104 (1997). If reproduction were not a major life activity, they argue, then it would have made little sense to include the reproductive disorders in the roster of physical impairments. This argument is simply wrong. There are numerous disorders of the reproductive system, such as dysmenorrhea and endo-metriosis, which are so painful that they limit a woman’s ability to engage in major life activities such as walking and working. And, obviously, cancer of the various reproductive organs limits one’s ability to engage in numerous activities other than reproduction.

But even if I were to assume that reproduction is a major life activity of respondent, I do not agree that an asymptomatic HIV infection “substantially limits” that activity. The record before us leaves no doubt that those so infected are still entirely able to engage in sexual intercourse, give birth *661to a child if they become pregnant, and perform the manual tasks necessary to rear a child to maturity. See App. 53-54. While individuals infected with HIV may choose not to engage in these activities, there is no support in language, logic, or our case law for the proposition that such voluntary choices constitute a “limit” on one’s own life activities.

The Court responds that the ADA “addresses substantial limitations on major life activities, not utter inabilities.” Ante, at 641. I agree, but fail to see how this assists the Court’s cause. Apart from being unable to demonstrate that she is utterly unable to engage in the various activities that comprise the reproductive process, respondent has not even explained how she is less able to engage in those activities.

Respondent contends that her ability to reproduce is limited because “the fatal nature of HIV infection means that a parent is unlikely to live long enough to raise and nurture the child to adulthood.” Brief for Respondent Abbott 22. But the ADA’s definition of a disability is met only if the alleged impairment substantially “limits” (present tense) a major life activity. 42 U. S. C. § 12102(2)(A). Asymptomatic HIV does not presently limit respondent’s ability to perform any of the tasks necessary to bear or raise a child. Respondent’s argument, taken to its logical extreme, would render every individual with a genetic marker for some debilitating disease “disabled” here and now because of some possible future effects.

In my view, therefore, respondent has failed to demonstrate that any of her major life activities were substantially limited by her HIV infection.

II

While the Court concludes to the contrary as to the “disability” issue, it then quite correctly recognizes that petitioner could nonetheless have refused to treat respondent if her condition posed a “direct threat.” The Court of Appeals *662affirmed the judgment of the District Court granting summary judgment to respondent on this issue. The Court vacates this portion of the Court of Appeals’ decision, and remands the case to the lower court, presumably so that it may “determine whether our analysis of some of the studies cited by the parties would change its conclusion that petitioner presented neither objective evidence nor a triable issue of fact on the question of risk.” Ante, at 655. I agree that the judgment should be vacated, although I am not sure I understand the Court’s cryptic direction to the lower court.

“[Djireet threat” is defined as a “significant risk to the health or safety of others that cannot be eliminated by a modification of policies, practices, or procedures or by the provision of auxiliary aids or services.” § 12182(b)(3). This statutory definition of a direct threat consists of two parts. First, a court must ask whether treating the infected patient without precautionary techniques would pose a “significant risk to the health or safety of others.” Ibid. Whether a particular risk is significant depends on:

“ '(a) the nature of the risk (how the disease is transmitted), (b) the duration of the risk (how long is the carrier infectious), (e) the severity of the risk (what is the potential harm to third parties) and (d) the probabilities the disease will be transmitted and will cause varying degrees of harm.”' School Bd. of Nassau Cty. v. Arline, 480 U. S. 273, 288 (1987).

Even if a significant risk exists, a health practitioner will still be required to treat the infected patient if “a modification of policies, practices, or procedures” (in this case, universal precautions) will “eliminat[ej” the risk. § 12182(b)(3).

' I agree with the Court that “[tjhe existence, or nonexistence, of a significant risk must be determined from the standpoint of the person who refuses the treatment or accommodation,” as of the time that the decision refusing treatment is made. Ante, at 649. I disagree with the Court, however, *663that “[i]n assessing the reasonableness of petitioner’s actions, the views of public health authorities ... are of special weight and authority.” Ante, at 650. Those views are, of course, entitled to a presumption of validity when the actions of those authorities themselves are challenged in court, and even in disputes between private parties where Congress has committed that dispute to adjudication by a public health authority. But in litigation between private parties originating in the federal courts, I am aware of no. provision of law or judicial practice that would require or permit courts to give some scientific views more credence than others simply because they have been endorsed by a politically appointed public health authority (such as the Surgeon General). In litigation of this latter sort, which is what we face here, the credentials of the scientists employed by the public health authority, and the soundness of their studies, must stand on their own. The Court cites no authority for its limitation upon the courts’ truth-finding function, except the statement in School Bd. of Nassau Cty. v. Arline, 480 U. S., at 288, that in making findings regarding the risk of contagion under the Rehabilitation Act, “courts normally should defer to the reasonable medical judgments of public health officials.” But there is appended to that dictum the following footnote, which makes it very clear that the Court was urging respect for medical judgment, and not necessarily respect for “official” medical judgment over “private” medical judgment: “This ease does not present, and we do not address, the question whether courts should also defer to the reasonable medical judgments of private physicians on which an employer has relied.” Id., at 288, n. 18.

Applying these principles here, it is clear to me that petitioner has presented more than enough evidence to avoid summary judgment on the “direct threat” question. In June 1994, the Centers for Disease Control and Prevention published a study identifying seven instances of possible transmission of HIV from patients to dental workers. See ante, *664at 654. While it is not entirely certain whether these dental workers contracted HIV during the course of providing dental treatment, the potential that the disease was transmitted during the course of dental treatment is relevant evidence. One need only demonstrate “risk,” not certainty of infection. See Arline, supra, at 288 (“ ‘[T]he probabilities the disease will be transmitted’ ” is a factor in assessing risk). Given the “severity of the risk” involved here, i. e., near certain death, and the fact that no public health authority had outlined a protocol for eliminating this risk in the context of routine dental treatment, it seems likely that petitioner ean establish that it was objectively reasonable for him to conclude that treating respondent in his office posed a “direct threat” to his safety.

In addition, petitioner offered evidence of 42 documented incidents of occupational transmission of HIV to health care workers other than dental professionals. App. 106. The Court of Appeals dismissed this evidence as irrelevant because these health professionals were not dentists. 107 F. 3d 934, 947 (CA1 1997). But the fact that the health care workers were not dentists is no more valid a basis for distinguishing these transmissions of HIV than the fact that the health care workers did not practice in Maine. At a minimum, petitioner’s evidence was sufficient to create a triable issue on this question, and summary judgment was accordingly not appropriate.

Justice O’Connor,

concurring in the judgment in part and dissenting in part.

I agree with The Chief Justice that respondent’s claim of disability should be evaluated on an individualized basis and that she has not proved that her asymptomatic HIV status substantially limited one or more of her major life activities. In my view, the act of giving birth to a child, while a very important part of the lives of many women, is not generally the same as the representative major life activities of *665all persons — “caring for one’s self, performing manual tasks, walking, seeing, hearing, speaking, breathing, learning, and working” — listed in regulations relevant to the Americans with Disabilities Act of 1990. See 45 CFR §84.3(j)(2)(ii) (1997); 28 CFR § 41.31(b)(2) (1997). Based on that conclusion, there is no need to address whether other aspects of intimate or family relationships not raised in this ease could constitute major life activities; nor is there reason to consider Whether HIV status would impose a substantial limitation on one’s ability to reproduce if reproduction were a major life activity.

I join in Part II of The Chief Justice’s opinion concurring in the judgment in part and dissenting in part, which concludes that the Court of Appeals failed to properly determine whether respondent’s condition posed a direct threat. Accordingly, I agree that a remand is necessary on that issue.