with whom Justice Brennan joins, dissenting.
The issue in this case is not whether declaring a war on illegal drugs is good public policy. The importance of ridding our society of such drugs is, by now, apparent to all. Rather, the issue here is whether the Government’s deployment in that war of a particularly Draconian weapon — the compulsory collection and chemical testing of railroad workers’ blood and urine — comports with the Fourth Amendment. Precisely because the need for action against the drug scourge is manifest, the need for vigilance against unconstitutional excess is great. History teaches that grave threats to liberty often come in times of urgency, when constitutional rights seem too extravagant to endure. The World War II relocation-camp cases, Hirabayashi v. United States, 320 U. S. 81 (1943); Korematsu v. United States, 323 U. S. 214 (1944), and the Red scare and McCarthy-era internal subversion cases, Schenck v. United States, 249 U. S. 47 (1919); Dennis v. United States, 341 U. S. 494 (1951), are only the most extreme reminders that when we allow fundamental freedoms to be sacrificed in the name of real or perceived exigency, we invariably come to regret it.
In permitting the Government to force entire railroad crews to submit to invasive blood and urine tests, even when it lacks any evidence of drug or alcohol use or other wrongdoing, the majority today joins those shortsighted courts which have allowed basic constitutional rights to fall prey to momentary emergencies. The majority holds that the need of the Federal Railroad Administration (FRA) to deter and diagnose train accidents outweighs any “minimal” intrusions on personal dignity and privacy posed by mass toxicological testing of persons who have given no indication whatsoever of *636impairment. Ante, at 624. In reaching this result, the majority ignores the text and doctrinal history of the Fourth Amendment, which require that highly intrusive searches of this type be based on probable cause, not on the evanescent cost-benefit calculations of agencies or judges. But the majority errs even under its own utilitarian standards, trivializing the raw intrusiveness of, and overlooking serious conceptual and operational flaws in, the FRA’s testing program. These flaws cast grave doubts on whether that program, though born of good intentions, will do more than ineffectually symbolize the Government’s opposition to drug use.
The majority purports to limit its decision to postaccident testing of workers in “safety-sensitive” jobs, ante, at 620, much as it limits its holding in the companion case to the testing of transferees to jobs involving drug interdiction or the use of firearms. Treasury Employees v. Von Raab, post, at 664. But the damage done to the Fourth Amendment is not so easily cabined. The majority’s acceptance of dragnet blood and urine testing ensures that the first, and worst, casualty of the war on drugs will be the precious liberties of our citizens. I therefore dissent.
I
The Court today takes its longest step yet toward reading the probable-cause requirement out of the Fourth Amendment. For the fourth time in as many years, a majority holds that a “‘special nee[d], beyond the normal need for law enforcement,’” makes the “‘requirement’” of probable cause “‘impracticable.’” Ante, at 619 (citations omitted). With the recognition of “[t]he Government’s interest in regulating the conduct of railroad employees to ensure safety” as such a need, ante, at 620, the Court has now permitted “special needs” to displace constitutional text in each of the four categories of searches enumerated in the Fourth Amendment: searches of “persons,” ante, at 613-614; “houses,” Griffin v. Wisconsin, 483 U. S. 868 (1987); “papers,” O’Connor v. Or*637tega, 480 U. S. 709 (1987); and “effects,” New Jersey v. T. L. O., 469 U. S. 325 (1985).
The process by which a constitutional “requirement” can be dispensed with as “impracticable” is an elusive one to me. The Fourth Amendment provides that “[t]he right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures, shall not be violated; and no Warrants shall issue, but upon probable cause, supported by Oath or affirmation, and particularly describing the place to be searched, and the persons or things to be seized.” The majority’s recitation of the Amendment, remarkably, leaves off after the word “violated,” ante, at 613, but the remainder of the Amendment — the Warrant Clause — is not so easily excised. As this Court has long recognized, the Framers intended the provisions of that Clause — a warrant and probable cause — to “provide the yardstick against which official searches and seizures are to be measured.” T. L. O., supra, at 359-360 (opinion of Brennan, J.). Without the content which those provisions give to the Fourth Amendment’s overarching command that searches and seizures be “reasonable,” the Amendment lies virtually devoid of meaning, subject to whatever content shifting judicial majorities, concerned about the problems of the day, choose to give to that supple term. See Dunaway v. New York, 442 U. S. 200, 213 (1979) (“[T]he protections intended by the Framers could all too easily disappear in the consideration and balancing of the multifarious circumstances presented by different cases”). Constitutional requirements like probable cause are not fair-weather friends, present when advantageous, conveniently absent when “special needs” make them seem not.
Until recently, an unbroken line of cases had recognized probable cause as an indispensable prerequisite for a full-scale search, regardless of whether such a search was conducted pursuant to a warrant or under one of the recognized exceptions to the warrant requirement. T. L. O., supra, at 358 *638and 359, n. 3 (opinion of Brennan, J.); see also Chambers v. Maroney, 399 U. S. 42, 51 (1970). Only where the government action in question had a “substantially less intrusive” impact on privacy, Dunaway, 442 U. S., at 210, and thus clearly fell short of a full-scale search, did we relax the probable-cause standard. Id., at 214 (“For all but those narrowly defined intrusions, the requisite ‘balancing’ ... is embodied in the principle that seizures are ‘reasonable’ only if supported by probable cause”); see also T. L. O., supra, at 360 (opinion of Brennan, J.). Even in this class of cases, we almost always required the government to show some individualized suspicion to justify the search.1 The few searches which we upheld in the absence of individualized justification were routinized, fleeting, and nonintrusive encounters conducted pursuant to regulatory programs which entailed no contact with the person.2
*639In the four years since this Court, in T. L. O., first began recognizing “special needs” exceptions to the Fourth Amendment, the clarity of Fourth Amendment doctrine has been badly distorted, as the Court has eclipsed the probable-cause requirement in a patchwork quilt of settings: public school principals’ searches of students’ belongings, T. L. O.; public employers’ searches of employees’ desks, O’Connor; and probation officers’ searches of probationers’ homes, Griffin,3 Tellingly, each time the Court has found that “special needs” counseled ignoring the literal requirements of the Fourth Amendment for such full-scale searches in favor of a formless and unguided “reasonableness” balancing inquiry, it has concluded that the search in question satisfied that test. I have joined dissenting opinions in each of these cases, protesting the “jettison[ing of] . . . the only standard that finds support in the text of the Fourth Amendment” and predicting that the majority’s “Rohrsehach-like ‘balancing test’” portended “a dangerous weakening of the purpose of the Fourth Amendment to protect the privacy and security of our citizens.” T. L. O., supra, at 357-358 (opinion of Brennan, J.).
The majority’s decision today bears out that prophecy. After determining that the Fourth Amendment applies to the FRA’s testing regime, the majority embarks on an extended inquiry into whether that regime is “reasonable,” an inquiry in which it balances “‘all of the circumstances surrounding the search or seizure and the nature of the search or seizure itself.’” Ante, at 619, quoting United States v. Montoya de *640Hernandez, 473 U. S. 531, 537 (1985). The result is “special needs” balancing analysis’ deepest incursion yet into the core protections of the Fourth Amendment. Until today, it was conceivable that, when a government search was aimed at a person and not simply the person’s possessions, balancing analysis had no place. No longer: with nary a word of explanation or acknowledgment of the novelty of its approach, the majority extends the “special needs” framework to a regulation involving compulsory blood withdrawal and urinary excretion, and chemical testing of the bodily fluids collected through these procedures. And until today, it was conceivable that a prerequisite for surviving “special needs” analysis was the existence of individualized suspicion. No longer: in contrast to the searches in T. L. O., O’Connor, and Griffin, which were supported by individualized evidence suggesting the culpability of the persons whose property was searched,4 the regulatory regime upheld today requires the postaccident collection and testing of the blood and urine of all covered employees — even if every member of this group gives every indication of sobriety and attentiveness.
In widening the “special needs” exception to probable cause to authorize searches of the human body unsupported by any evidence of wrongdoing, the majority today completes the process begun in T. L. O. of eliminating altogether the probable-cause requirement for civil searches — those undertaken for reasons “beyond the normal need for law enforcement.” Ante, at 619 (citations omitted). In its place, the majority substitutes a manipulable balancing inquiry under which, upon the mere assertion of a “special need,” even the deepest dignitary and privacy interests become vul*641nerable to governmental incursion. See ibid, (distinguishing criminal from civil searches). By its terms, however, the Fourth Amendment — unlike the Fifth and Sixth — does not confine its protections to either criminal or civil actions. Instead, it protects generally “[t]he right of the people to be secure.”5
The fact is that the malleable “special needs” balancing approach can be justified only on the basis of the policy results it allows the majority to reach. The majority’s concern with the railroad safety problems caused by drug and alcohol abuse is laudable; its cavalier disregard for the text of the Constitution is not. There is no drug exception to the Constitution, any more than there is a communism exception or an exception for other real or imagined sources of domestic unrest. Coolidge v. New Hampshire, 403 U. S. 443, 455 (1971). Because abandoning the explicit protections of the Fourth Amendment seriously imperils “the right to be let alone — the most comprehensive of rights and the right most valued by civilized men,” Olmstead v. United States, 277 U. S. 438, 478 (1928) (Brandeis, J., dissenting), I reject the majority’s “special needs” rationale as unprincipled and dangerous.
II
The proper way to evaluate the FRA’s testing regime is to use the same analytic framework which we have traditionally used to appraise Fourth Amendment claims involving full-scale searches, at least until the recent “special needs” cases. Under that framework, we inquire, serially, whether a *642search has taken place, see, e. g., Katz v. United States, 389 U. S. 347, 350-353 (1967); whether the search was based on a valid warrant or undertaken pursuant to a recognized exception to the warrant requirement, see, e. g., Welsh v. Wisconsin, 466 U. S. 740, 748-750 (1984); whether the search was based on probable cause or validly based on lesser suspicion because it was minimally intrusive, see, e. g., Dunaway, 442 U. S., at 208-210; and, finally, whether the search was conducted in a reasonable manner, see, e. g., Winston v. Lee, 470 U. S. 753, 763-766 (1985). See also T. L. O., 469 U. S., at 354-355 (opinion of Brennan, J.) (summarizing analytic framework).
The majority’s threshold determination that “covered” railroad employees have been searched under the FRA’s testing program is certainly correct. Ante, at 616-618. Who among us is not prepared to consider reasonable a person’s expectation of privacy with respect to the extraction of his blood, the collection of his urine, or the chemical testing of these fluids? United States v. Jacobsen, 466 U. S. 109, 113 (1984).6 The majority’s ensuing conclusion that the warrant requirement may be dispensed with, however, conveniently overlooks the fact that there are three distinct searches at issue. Although the importance of collecting blood and urine samples before drug or alcohol metabolites disappear justifies waiving the warrant requirement for those two searches under the narrow “exigent circumstances” exception, see Schmerber v. California, 384 U. S. 757, 770 (1966) (“[T]he delay necessary to obtain a warrant. . . threatens] ‘the destruction of evidence’ ”), no such exigency prevents railroad officials from securing a warrant before chemically testing the samples they obtain. Blood and urine do not spoil if *643properly collected and preserved, and there is no reason to doubt the ability of railroad officials to grasp the relatively simple procedure of obtaining a warrant authorizing, where appropriate, chemical analysis of the extracted fluids. It is therefore wholly unjustified to dispense with the warrant requirement for this final search. See Chimel v. California, 395 U. S. 752, 761-764 (1969) (exigency exception permits warrantless searches only to the extent that exigency exists).
It is the probable-cause requirement, however, that the FRA’s testing regime most egregiously violates, a fact which explains the majority’s ready acceptance and expansion of the countertextual “special needs” exception. By any measure, the FRA’s highly intrusive collection and testing procedures qualify as full-scale personal searches. Under our precedents, a showing of probable cause is therefore clearly required. But even if these searches were viewed as entailing only minimal intrusions on the order, say, of a police stop- and-frisk, the FRA’s program would still fail to pass constitutional muster, for we have, without exception, demanded that even minimally intrusive searches of the person be founded on individualized suspicion. See supra, at 638, and n. 1. The federal parties concede it does not satisfy this standard. Brief for Federal Parties 18. Only if one construes the FRA’s collection and testing procedures as akin to the routinized and fleeting regulatory interactions which we have permitted in the absence of individualized suspicion, see n. 2, supra, might these procedures survive constitutional scrutiny. Presumably for this reason, the majority likens this case to United States v. Martinez-Fuerte, 428 U. S. 543 (1976), which upheld brief automobile stops, at the border to ascertain the validity of motorists’ residence in the United States. Ante, at 624. CaseTaw and common sense reveal both the bankruptcy of this absurd analogy and the constitutional imperative of adhering to the textual standard of probable cause to evaluate the FRA’s multifarious full-scale searches.
*644Compelling a person to submit to the piercing of his skin by a hypodermic needle so that his blood may be extracted significantly intrudes on the “personal privacy and dignity against unwarranted intrusion by the State” against which the Fourth Amendment protects. Schmerber, supra, at 767. As we emphasized in Terry v. Ohio, 392 U. S. 1, 24-25 (1968), “Even a limited search of the outer clothing . . . constitutes a severe, though brief, intrusion upon cherished personal security, and it must surely be an annoying, frightening, and perhaps humiliating experience.” We have similarly described the taking of a suspect’s fingernail scrapings as a “ ‘severe, though brief, intrusion upon cherished personal security.’” Cupp v. Murphy, 412 U. S. 291, 295 (1973) (quoting Terry, supra, at 24-25, and upholding this procedure upon a showing of probable cause). The government-compelled withdrawal of blood, involving as it does the added aspect of physical invasion, is surely no less an intrusion. The surrender of blood on demand is, furthermore, hardly a quotidian occurrence. Cf. Martinez-Fuerte, supra, at 557 (routine stops involve “quite limited” intrusion).
In recognition of the intrusiveness of this procedure, we specifically required in Schmerber that police have evidence of a drunken-driving suspect’s impairment before forcing him to endure a blood test:
“The interests in human dignity and privacy which the Fourth Amendment protects forbid any such intrusions on the mere chance that desired evidence might be obtained. In the absence of a clear indication that in fact such evidence will be found, these fundamental human interests require law officers to suffer the risk that such evidence may disappear . . . .” 384 U. S., at 769-770.
Schmerber strongly suggested that the “clear indication” needed to justify a compulsory blood test amounted to a showing of probable cause, which “plainly” existed in that case. Id., at 768. Although subsequent cases interpreting Schmerber have differed over whether a showing of individ*645ualized suspicion would have sufficed, compare Winston, 470 U. S., at 760 (Schmerber “noted the importance of probable cause”), with Montoya de Hernandez, 473 U. S., at 540 (Schmerber “indicate[d] the necessity for particularized suspicion”), by any reading, Schmerber clearly forbade compulsory blood tests on any lesser showing than individualized suspicion. Exactly why a blood test which, if conducted on one person, requires a showing of at least individualized suspicion may, if conducted on many persons, be based on no showing whatsoever, the majority does not — and cannot — explain.7
Compelling a person to produce a urine sample on demand also intrudes deeply on privacy and bodily integrity. Urination is among the most private of activities. It is generally forbidden in public, eschewed as a matter of conversation, and performed in places designed to preserve this tradition of *646personal seclusion. Cf. Martinez-Fuerte, 428 U. S., at 560 (border-stop questioning involves no more than “some annoyance” and is neither “frightening” nor “offensive”). The FRA, however, gives scant regard to personal privacy, for its Field Manual instructs supervisors monitoring urination that railroad workers must provide urine samples “under direct observation by the physician/technician.” Federal Railroad Administration, United States Dept, of Transportation, Field Manual: Control of Alcohol and Drug Use in Railroad Operations D-5 (1986) (emphasis added).8 That the privacy interests offended by compulsory and supervised urine collection are profound is the overwhelming judgment of the lower courts and commentators. As Professor — later Solicitor General — Charles Fried has written:
“[I]n our culture the excretory functions are shielded by more or less absolute privacy, so much so that situations in which this privacy is violated are experienced as extremely distressing, as detracting from one’s dignity and self esteem.” Privacy, 77 Yale L. J. 475, 487 (1968).9
The majority’s characterization of the privacy interests implicated by urine collection as “minimal,” ante, at 624, is noth*647ing short of startling. This characterization is, furthermore, belied by the majority’s own prior explanation of why compulsory urination constitutes a search for the purposes of the Fourth Amendment:
“ ‘There are few activities in our society more personal or private than the passing of urine. Most people describe it by euphemisms if they talk about it at all. It is a function traditionally performed without public observation; indeed, its performance in public is generally prohibited by law as well as social custom.’” Ante, at 617, quoting National Treasury Employees Union v. Von Raab, 816 F. 2d 170, 175 (CA5 1987).
The fact that the majority can invoke this powerful passage in the context of deciding that a search has occurred, and then ignore it in deciding that the privacy interests this search implicates are “minimal,” underscores the shameless manipulability of its balancing approach.
Finally, the chemical analysis the FRA performs upon the blood and urine samples implicates strong privacy interests apart from those intruded upon by the collection of bodily fluids. Technological advances have made it possible to uncover, through analysis of chemical compounds in these fluids, not only drug or alcohol use, but also medical disorders such as epilepsy, diabetes, and clinical depression. Cf. Martinez-Fuerte, supra, at 558, quoting United States v. Brignoni-Ponce, 422 U. S. 873, 880 (1975) (checkpoint inquiry involves only “ ‘a brief question or two’ ” about motorist’s residence). As the Court of Appeals for the District of Columbia Circuit has observed: “[S]uch tests may provide Government officials with a periscope through which they can peer into an individual’s behavior in her private life, even in her own home.” Jones v. McKenzie, 266 U. S. App. D. C. 85, 89, 833 F. 2d 335, 339 (1987); see also Capua v. Plainfield, 643 F. Supp. 1507, 1511 (NJ 1986) (urine testing is “form of surveillance” which “reports on a person’s off-duty activities just as surely as someone had been present and *648watching”)- The FRA’s requirement that workers disclose the medications they have taken during the 30 days prior to chemical testing further impinges upon the confidentiality customarily attending personal health secrets.
By any reading of our precedents, the intrusiveness of these three searches demands that they — like other full-scale searches — be justified by probable cause. It is no answer to suggest, as does the majority, that railroad workers have relinquished the protection afforded them by this Fourth Amendment requirement, either by “participating] in an industry that is regulated pervasively to ensure safety” or by undergoing periodic fitness tests pursuant to state law or to collective-bargaining agreements. Ante, at 627.
Our decisions in the regulatory search area refute the suggestion that the heavy regulation of the railroad industry eclipses workers’ rights under the Fourth Amendment to insist upon a showing of probable cause when their bodily fluids are being extracted. This line of cases has exclusively involved searches of employer property, with respect to which “[c]ertain industries have such a history of government oversight that no reasonable expectation of privacy could exist for a proprietor over the stock of such an enterprise.” Marshall v. Barlow’s, Inc., 436 U. S. 307, 313 (1978) (emphasis added; citation omitted), quoted in New York v. Burger, 482 U. S. 691, 700 (1987). Never have we intimated that regulatory searches reduce employees’ rights of privacy in their persons. See Camara v. Municipal Court of San Francisco, 387 U. S. 523, 537 (1967) (“[T]he inspections are [not] personal in nature”); cf. Donovan v. Dewey, 452 U. S. 594, 598-599 (1981); Marshall, supra, at 313. As the Court pointed out in O’Con-nor, individuals do not lose Fourth Amendment rights at the workplace gate, 480 U. S., at 716-718; see also Oliver v. United States, 466 U. S. 170, 178, n. 8 (1984), any more than they relinquish these rights at the schoolhouse door, T. L. O., 469 U. S., at 333, or the hotel room threshold, Hoff a v. United States, 385 U. S. 293, 301 (1966). These rights mean *649little indeed if, having passed through these portals, an individual may remain subject to a suspicionless search of his person justified solely on the grounds that the government already is permitted to conduct a search of the inanimate contents of the surrounding area. In holding that searches of persons may fall within the category of regulatory searches permitted in the absence of probable cause or even individualized suspicion, the majority sets a dangerous and ill-conceived precedent.
The majority’s suggestion that railroad workers’ privacy is only minimally invaded by the collection and testing of their bodily fluids because they undergo periodic fitness tests, ante, at 624-625, is equally baseless. As an initial matter, even if participation in these fitness tests did render “minimal” an employee’s “interest in bodily security,” ante, at 628, such minimally intrusive searches of the person require, under our precedents, a justificatory showing of individualized suspicion. See supra, at 637. More fundamentally, railroad employees are not routinely required to submit to blood or urine tests to gain or to maintain employment, and railroad employers do not ordinarily have access to employees’ blood or urine, and certainly not for the purpose of ascertaining drug or alcohol usage. That railroad employees sometimes undergo tests of eyesight, hearing, skill, intelligence, and agility, ante, at 627, n. 8, hardly prepares them for Government demands to submit to the extraction of blood, to excrete under supervision, or to have these bodily fluids tested for the physiological and psychological secrets they may contain. Surely employees who release basic information about their financial and personal history so that employers may ascertain their “ethical fitness” do not, by so doing, relinquish their expectations of privacy with respect to their personal letters and diaries, revealing though these papers may be of their character.
I recognize that invalidating the full-scale searches involved in the FRA’s testing regime for failure to comport with the Fourth Amendment’s command of probable cause *650may hinder the Government’s attempts to make rail transit as safe as humanly possible. But constitutional rights have their consequences, and one is that efforts to maximize the public welfare, no matter how well intentioned, must always be pursued within constitutional boundaries. Were the police freed from the constraints of the Fourth Amendment for just one day to seek out evidence of criminal wrongdoing, the resulting convictions and incarcerations would probably prevent thousands of fatalities. Our refusal to tolerate this specter reflects our shared belief that even beneficent governmental power — whether exercised to save money, save lives, or make the trains run on time — must always yield to “a resolute loyalty to constitutional safeguards.” Almeida-Sanchez v. United States, 413 U. S. 266, 273 (1973). The Constitution demands no less loyalty here.
I — I K — I hH
Even accepting the majority’s view that the FRA s collection and testing program is appropriately analyzed under a multifactor balancing test, and not under the literal terms of the Fourth Amendment, I would still find the program invalid. The benefits of suspicionless blood and urine testing are far outstripped by the costs imposed on personal liberty by such sweeping searches. Only by erroneously deriding as “minimal” the privacy and dignity interests at stake, and by uncritically inflating the likely efficacy of the FRA’s testing program, does the majority strike a different balance.
For the reasons stated above, I find nothing minimal about the intrusion on individual liberty that occurs whenever the Government forcibly draws and analyzes a person’s blood and urine. Several aspects of the FRA’s testing program exacerbate the intrusiveness of these procedures. Most strikingly, the agency’s regulations not only do not forbid, but, in fact, appear to invite criminal prosecutors to obtain the blood and urine samples drawn by the FRA and use them as the basis of criminal investigations and trials. See 49 CFR *651§ 219.211(d) (1987) (“Each sample . . . may be made available to ... a party in litigation upon service of appropriate compulsory process on the custodian of the sample . . .”). This is an unprecedented invitation, leaving open the possibility of criminal prosecutions based on suspicionless searches of the human body. Cf. Treasury Employees, post, at 666 (Customs Service drug-testing program prohibits use of test results in criminal prosecutions); Camara, 387 U. S., at 537.
Tobe sure, the majority acknowledges, in passing, the possibility of criminal prosecutions, ante, at 621, n. 5-, but it refuses to factor this possibility into its Fourth Amendment balancing process, stating that “the record does not disclose that [49 CFR § 219.211(d) (1987)] was intended to be, or actually has been, so used.” Ibid. This demurrer is highly disingenuous. The federal parties concede that they find “no prohibition on the release of FRA testing results to prosecutors.” Brief for Federal Parties 10, n. 15. The absence of prosecutions to date — which is likely due to the fact that the FRA’s regulations have been held invalid for much of their brief history — hardly proves that prosecutors will not avail themselves of the FRA’s invitation in the future. If the majority really views the impact of FRA testing on privacy interests as minimal even if these tests generate criminal prosecutions, it should say so. If the prospect of prosecutions would lead the majority to reassess the validity of the testing program with prosecutions as part of the balance, it should say so, too, or condition its approval of that program on the nonrelease of test results to prosecutors. In ducking this important issue, the majority gravely disserves both the values served by the Fourth Amendment and the rights of those persons whom the FRA searches. Furthermore, the majority’s refusal to restrict the release of test results casts considerable doubt on the conceptual basis of its decision— that the “special need” of railway safety is one “beyond the *652normal need for law enforcement.” Ante, at 619 (citations omitted).10
The majority also overlooks needlessly intrusive aspects of the testing process itself. Although the FRA requires the collection and testing of both blood and urine, the agency concedes that mandatory urine tests — unlike blood tests — do not measure current impairment and therefore cannot differentiate on-duty impairment from prior drug or alcohol use which has ceased to affect the user’s behavior. See 49 CFR § 219.309(2) (1987) (urine test may reveal use of drugs or alcohol as much as 60 days prior to sampling). Given that the FRA’s stated goal is to ascertain current impairment, and not to identify persons who have used substances in their spare time sufficiently in advance of their railroad duties to pose no risk of on-duty impairment, §219.101(a), mandatory urine testing seems wholly excessive. At the very least, the FRA could limit its use of urinalysis to confirming findings of current impairment suggested by a person’s blood tests. The additional invasion caused by automatically testing urine as well as blood hardly ensures that privacy interests “will be invaded no more than is necessary.” T. L. O., 469 U. S., at 343.
The majority’s trivialization of the intrusions on worker privacy posed by the FRA’s testing program is matched at the other extreme by its blind acceptance of the Government’s assertion that testing will “dete[r] employees engaged in safety-sensitive tasks from using controlled substances or alcohol,” and “help railroads obtain invaluable information *653about the causes of major accidents.” Ante, at 629, 630. With respect, first, to deterrence, it is simply implausible that testing employees after major accidents occur, 49 CFR § 219.201(a)(1) (1987), will appreciably discourage them from using drugs or alcohol. As Justice Stevens observes in his concurring opinion:
“Most people — and I would think most railroad employees as well — do not go to work with the expectation that they may be involved in a major accident, particularly one causing such catastrophic results as loss of life or the release of hazardous material requiring an evacuation. Moreover, even if they are conscious of the possibilities that such an accident might occur and that alcohol or drug use might be a contributing factor, if the risk of serious personal injury does not deter their use of these substances, it seems highly unlikely that the additional threat of loss of employment would have any effect on their behavior.” Ante, at 634.
Under the majority’s deterrence rationale, people who skip school or work to spend a sunny day at the zoo will not taunt the lions because their truancy or absenteeism might be discovered in the event they are mauled. It is, of course, the fear of the accident, not the fear of a postaccident revelation, that deters. The majority’s credulous acceptance of the FRA’s deterrence rationale is made all the more suspect by the agency’s failure to introduce, in an otherwise ample administrative record, any studies explaining or supporting its theory of accident deterrence.
The poverty of the majority’s deterrence rationale leaves the Government’s interest in diagnosing the causes of major accidents as the sole remaining justification for the FRA’s testing program. I do not denigrate this interest, but it seems a slender thread from which to hang such an intrusive program, particularly given that the knowledge that one or more workers were impaired at the time of an accident falls far short of proving that substance abuse caused or exacer*654bated that accident. See 839 F. 2d 575, 587 (CA9 1988). Some corroborative evidence is needed: witness or co-worker accounts of a worker’s misfeasance, or at least indications that the cause of the accident was within a worker’s area of responsibility. Such particularized facts are, of course, the very essence of the individualized suspicion requirement which the respondent railroad workers urge, and which the Court of Appeals found to “pos[e] no insuperable burden on the government.” Id., at 588. Furthermore, reliance on the importance of diagnosing the causes of an accident as a critical basis for upholding the FRA’s testing plan is especially hard to square with our frequent admonition that the interest in ascertaining the causes of a criminal episode does not justify departure from the Fourth Amendment’s requirements. “[T]his Court has never sustained a search upon the sole ground that officers reasonably expected to find evidence of a particular crime . . . .” Katz, 389 U. S., at 356. Nor should it here.
IV
In his first dissenting opinion as a Member of this Court, Oliver Wendell Holmes observed:
“Great cases, like hard cases, make bad law. For great cases are called great, not by reason of their real importance in shaping the law of the future, but because of some accident of immediate overwhelming interest which appeals to the feelings and distorts the judgment. These immediate interests exercise a kind of hydraulic pressure which makes what previously was clear seem doubtful, and before which even well settled principles of law will bend.” Northern Securities Co. v. United States, 193 U. S. 197, 400-401 (1904).
A majority of this Court, swept away by society’s obsession with stopping the scourge of illegal drugs, today succumbs to the popular pressures described by Justice Holmes. In upholding the FRA’s plan for blood and urine testing, the *655majority bends time-honored and textually based principles of the Fourth Amendment — principles the Framers of the Bill of Rights designed to ensure that the Government has a strong and individualized justification when it seeks to invade an individual’s privacy. I believe the Framers would be appalled by the vision of mass governmental intrusions upon the integrity of the human body that the majority allows to become reality. The immediate victims of the majority’s constitutional timorousness will be those railroad workers whose bodily fluids the Government may now forcibly collect and analyze. But ultimately, today's decision will reduce the privacy all citizens may enjoy, for, as Justice Holmes understood, principles of law, once bent, do not snap back easily. I dissent.
The first, and leading, case of a minimally intrusive search held valid when based on suspicion short of probable cause is Terry v. Ohio, 392 U. S. 1, 30 (1968), where we held that a police officer who observes unusual conduct suggesting criminal activity by persons he reasonably suspects are armed and presently dangerous may “conduct a carefully limited search of the outer clothing of such persons.” See also United States v. Hensley, 469 U. S. 221 (1985) (upholding brief stop of person described on wanted flyer while police ascertain if arrest warrant has been issued); Delaware v. Prouse, 440 U. S. 648 (1979) (invalidating discretionary stops of motorists to check licenses and registrations when not based on reasonable suspicion that the motorist is unlicensed, the automobile is unregistered, or that the vehicle or an occupant should otherwise be detained); Pennsylvania v. Minims, 434 U. S. 106 (1977) (upholding limited search where officers who had lawfully stopped car saw a large bulge under the driver’s jacket); United States v. Brignoni-Ponce, 422 U. S. 873 (1975) (upholding brief stops by roving border patrols where officers reasonably believe car may contain illegal aliens); Adams v. Williams, 407 U. S. 143 (1972) (upholding brief stop to interrogate suspicious individual believed to be carrying narcotics and gun).
See, e. g., United States v. Martinez-Fuerte, 428 U. S. 543 (1976) (brief interrogative stop at permanent border checkpoint to ascertain motorist’s residence status); Camara v. Municipal Court of San Fran*639cisco, 387 U. S. 523 (1967) (routine annual inspection by city housing department).
The “special needs” the Court invoked to justify abrogating the probable-cause requirement were, in New Jersey v. T. L. O., 469 U. S., at 341, “the substantial need of teachers and administrators for freedom to maintain order in the schools”; in O’Connor v. Ortega, 480 U. S., at 725, “the efficient and proper operation of the workplace”; and in Griffin v. Wisconsin, 483 U. S., at 878, the need to preserve “the deterrent effect of the supervisory arrangement” of probation.
See T. L. O., supra, at 346 (teacher’s report that student had been smoking provided reasonable suspicion that purse contained cigarettes); O’Connor, supra, at 726 (charges of specific financial improprieties gave employer individualized suspicion of misconduct by employee); Griffin, supra, at 879-880 (tip to police officer that probationer was storing guns in his apartment provided reasonable suspicion).
That the Fourth Amendment applies equally to criminal and civil searches was emphasized, ironically enough, in the portion of T. L. O. holding the Fourth Amendment applicable to schoolhouse searches. 469 U. S., at 335. The malleability of “special needs” balancing thus could not be clearer: the majority endorses the applicability of the Fourth Amendment to civil searches in determining whether a search has taken place, but then wholly ignores it in the subsequent inquiry into the validity of that search.
The FRA’s breath-testing procedures also constitute searches subject to constitutional safeguards. See ante, at 616-617 (reaching same conclusion). I focus my discussion on the collection and testing of blood and urine because those more intrusive procedures better demonstrate the excesses of the FRA’s scheme.
The majority, seeking to lessen the devastating ramifications of Schmerber v. California, and to back up its assertion that Government-imposed blood extraction does not “infringe significant privacy interests,” ante, at 625, emphasizes Schmerber's observation that blood tests are commonplace and can be performed with “‘virtually no risk, trauma, or pain.’” Ante, at 625, quoting 384 U. S., at 771. The majority, however, wrenches this statement out of context. The Schmerber Court made this statement only after it established that the blood test fell within the “exigent circumstances” exception to the warrant requirement, and that the test was supported by probable cause. Indeed, the statement was made only in the context of the separate inquiry into whether the compulsory blood test was conducted in a reasonable manner. 384 U. S., at 768-772; see also Winston v. Lee, 470 U. S. 753, 760-761 (1985) (“Schmerber recognized that the ordinary requirements of the Fourth Amendment would be the threshold requirements for conducting this kind of surgical search and seizure. . . . Beyond these standards, Schmerber's inquiry considered a number of other factors in determining the ‘reasonableness’ of the blood test”) (emphasis added). The majority also cites South■ Dakota v. Neville, 459 U. S. 553 (1983), and Breithaupt v. Abram, 352 U. S. 432 (1957), for the proposition that blood tests are commonplace. Ante, at 625. In both those cases, however, the police officers who attempted to impose blood tests on drunken-driving suspects had exceptionally strong evidence of the driver’s inebriation. 459 U. S., at 554-556; 352 U. S., at 433.
The majority dismisses as nonexistent the intrusiveness of such “direct observation,” on the ground that FRA regulations state that such observation is not “require[d].” 50 Fed. Reg. 31555 (1985), cited ante, at 626. The majority’s dismissal is too hasty, however, for the regulations — in the very same sentence — go on to state: “[B]ut observation is the most effective means of ensuring that the sample is that of the employee and has not been diluted.” 50 Fed. Reg. 31555 (1985). Even if this were not the case, the majority’s suggestion that officials monitoring urination will disregard the clear commands of the Field Manual with which they are provided is dubious, to say the least.
See, e. g., National Treasury Employees Union v. Von Raab, 816 F. 2d 170, 175 (CA5 1987), aff’d in pertinent part, post, p. 656; Taylor v. O'Grady, 669 F. Supp. 1422, 1433-1434 (ND Ill. 1987); Feliciano v. Cleveland, 661 F. Supp. 578, 586 (ND Ohio 1987); American Federation of Government Employees, AFL-CIO v. Weinberger, 651 F. Supp. 726, 732-733 (SD Ga. 1986); Capua v. Plainfield, 643 F. Supp. 1507, 1514 (NJ 1986).
As a result of the majority’s extension of the regulatory search doctrine to searches of the person, individuals the FRA finds to have used drugs may face criminal prosecution, even if their impairment had nothing to do with causing an accident. The majority observes that evidence of criminal behavior unearthed during an otherwise valid regulatory search is not excludible unless the search is shown to be a “pretext” for obtaining evidence for a criminal trial, ante, at 621, n. 5, citing New York v. Burger, 482 U. S. 691, 716-717, n. 27 (1987) — a defense the majority belittles but, mercifully, preserves for another day.