with whom Justice Stevens, Justice O’Connor, and Justice Souter join, dissenting.
Seven years ago, in Vernonia School Dist. 47J v. Acton, 515 U. S. 646 (1995), this Court determined that a school *843district’s policy of randomly testing the urine of its student athletes for illicit drugs did not violate the Fourth Amendment. In so ruling, the Court emphasized that drug use “increase[d] the risk of sports-related injury” and that Ver-nonia’s athletes were the “leaders” of an aggressive local “drug culture” that had reached “‘epidemic proportions.’” Id., at 649. Today, the Court relies upon Vernonia to permit a school district with a drug problem its superintendent repeatedly described as “not. . . major,” see App. 180, 186, 191, to test the urine of an academic team member solely by reason of her participation in a nonathletic, competitive extracurricular activity — participation associated with neither special dangers from, nor particular predilections for, drug use.
“[T]he legality of a search of a student,” this Court has instructed, “should depend simply on the reasonableness, under all the circumstances, of the search.” New Jersey v. T. L. O., 469 U. S. 325, 341 (1985). Although “ ‘special needs’ inhere in the public school context,” see ante, at 829 (quoting Vernonia, 515 U. S., at 653), those needs are not so expansive or malleable as to render reasonable any program of student drug testing a school district elects to install. The particular testing program upheld today is not reasonable; it is capricious, even perverse: Petitioners’ policy targets for testing a student population least likely to be at risk from illicit drugs and their damaging effects. I therefore dissent.
1 — <
A
A search unsupported by probable cause nevertheless may be consistent with the Fourth Amendment “when special needs, beyond the normal need for law enforcement, make the warrant and probable-cause requirement impracticable.” Griffin v. Wisconsin, 483 U. S. 868, 873 (1987) (internal quotation marks omitted). In Vernonia, this Court made clear that “such ‘special needs’ . . . exist in the public school con*844text.” 515 U. S., at 653 (quoting Griffin, 483 U. S., at 873). The Court observed:
“[Wjhile children assuredly do not ‘shed their constitutional rights ... at the schoolhouse gate,’ Tinker v. Des Moines Independent Community School Dist., 393 U. S. 503, 506 (1969), the nature of those rights is what is appropriate for children in school.... Fourth Amendment rights, no less than First and Fourteenth Amendment rights, are different in public schools than elsewhere; the ‘reasonableness’ inquiry cannot disregard the schools’ custodial and tutelary responsibility for children.” 515 U. S., at 655-656 (other citations omitted).
The Vernonia Court concluded that a public school district facing a disruptive and explosive drug abuse problem sparked by members of its athletic teams had “special needs” that justified suspicionless testing of district athletes as a condition of their athletic participation.
This case presents circumstances dispositively different from those of Vernonia. True, as the Court stresses, Tecumseh students participating in competitive extracurricular activities other than athletics share two relevant characteristics with the athletes of Vernonia. First, both groups attend public schools. “[0]ur decision in Vernonia,” the Court states, “depended primarily upon the school’s custodial responsibility and authority.” Ante, at 831; see also ante, at 840 (Breyer, J., concurring) (school districts act in loco parentis). Concern for student health and safety is basic to the school’s caretaking, and it is undeniable that “drug use carries a variety of health risks for children, including death from overdose.” Ante, at 836-837 (majority opinion).
Those risks, however, are present for all schoolchildren. Vernonia cannot be read to endorse invasive and suspicion-less drug testing of all students upon any evidence of drug use, solely because drugs jeopardize the life and health of those who use them. Many children, like many adults, en*845gage in dangerous activities on their own time; that the children are enrolled in school scarcely allows government to monitor all such activities. If a student has a reasonable subjective expectation of privacy in the personal items she brings to school, see T. L. O., 469 U. S., at 338-339, surely she has a similar expectation regarding the chemical composition of her urine. Had the Vernonia Court agreed that public school attendance, in and of itself, permitted the State to test each student’s blood or urine for drugs, the opinion in Vernonia could have saved many words. See, e. g., 515 U. S., at 662 (“[I]t must not be lost sight of that [the Vernonia School District] program is directed ... to drug use by school athletes, where the risk of immediate physical harm to the drug user or those with whom he is playing his sport is particularly high.”).
The second commonality to which the Court points is the voluntary character of both interscholastic athletics and other competitive extracurricular activities. “By choosing to ‘go out for the team,’ [school athletes] voluntarily subject themselves to a degree of regulation even higher than that imposed on students generally.” Id., at 657. Comparably, the Court today observes, “students who participate in competitive extracurricular activities voluntarily subject themselves to” additional rules not applicable to other students. Ante, at 831.
The comparison is enlightening. While extracurricular activities are “voluntary” in the sense that they are not required for graduation, they are part of the school’s educational program; for that reason, the petitioner (hereinafter School District) is justified in expending public resources to ■ make them available. Participation in such activities is a key component of school life, essential in reality for students applying to college, and, for all participants, a significant contributor to the breadth and quality of the educational experience. See Brief for Respondents 6; Brief for American Academy of Pediatrics et al. as Amici Curiae 8-9. Students *846“volunteer” for extracurricular pursuits in the same way they might volunteer for honors classes: They subject themselves to additional requirements, but they do so in order to take full advantage of the education offered them. Cf. Lee v. Weisman, 505 U. S. 577, 595 (1992) (“Attendance may not be required by official decree, yet it is apparent that a student is not free to absent herself from the graduation exercise in any real sense of the term ‘voluntary,’ for absence would require forfeiture of those intangible benefits which have motivated the student through youth and all her high school years.”).
Voluntary participation in athletics has a distinctly different dimension: Schools regulate student athletes discretely because competitive school sports by their nature require communal undress and, more important, expose students to physical risks that schools have a duty to mitigate. For the very reason that schools cannot offer a program of competitive athletics without intimately affecting the privacy of students, Vernonia reasonably analogized school athletes to “adults who choose to participate in a closely regulated industry.” 515 U. S., at 657 (internal quotation marks omitted). Industries fall within the closely regulated category when the nature of their activities requires substantial government oversight. See, e. g., United States v. Biswell, 406 U. S. 311, 815-316 (1972). Interscholastic athletics similarly require close safety and health regulation; a school’s choir, band, and academic team do not.
In short, Vernonia applied, it did not repudiate, the principle that “the legality of a search of a student should depend simply on the reasonableness, under all the circumstances, of the search.” T. L. O., 469 U. S., at 341 (emphasis added). Enrollment in a public school, and election to participate in school activities beyond the bare minimum that the curriculum requires, are indeed factors relevant to reasonableness, but they do not on their own justify intrusive, suspicionless searches. Vernonia, accordingly, did not rest upon these *847factors; instead, the Court performed what today’s majority aptly describes as a “fact-specific balancing,” ante, at 830. Balancing of that order, applied to the facts now before the Court, should yield a result other than the one the Court announces today.
B
Vernonia initially considered “the nature of the privacy interest upon which the search [there] at issue intrude[d].” 515 U. S., at 654. The Court emphasized that student athletes’ expectations of privacy are necessarily attenuated:
“Legitimate privacy expectations are even less with regard to student athletes. School sports are not for the bashful. They require ‘suiting up’ before each practice or event, and showering and changing afterwards. Public school locker rooms, the usual sites for these activities, are not notable for the privacy they afford. The locker rooms in Vernonia are typical: No individual dressing rooms are provided; shower heads are lined up along a wall, unseparated by any sort of partition or curtain; not even all the toilet stalls have doors. . . . [T]here is an element of communal undress inherent in athletic participation.” Id., at 657 (internal quotation marks omitted).
Competitive extracurricular activities other than athletics, however, serve students of all manner: the modest and shy along with the bold and uninhibited. Activities of the kind plaintiff-respondent Lindsay Earls pursued — choir, show choir, marching band, and academic team — afford opportunities to gain self-assurance, to “come to know faculty members in a less formal setting than the typical classroom,” and to acquire “positive social supports and networks [that] play a critical role in periods of heightened stress.” Brief for American Academy of Pediatrics et al. as Amici Curiae 13.
On “occasional out-of-town trips,” students like Lindsay Earls “must sleep together in communal settings and use *848communal bathrooms.” 242 F. 3d 1264, 1275 (CA10 2001). But those situations are hardly equivalent to the routine communal undress associated with athletics; the School District itself admits that when such trips occur, “public-like restroom facilities,” which presumably include enclosed stalls, are ordinarily available for changing, and that “more modest students” find other ways to maintain their privacy. Brief for Petitioners 34.1
After describing school athletes’ reduced expectation of privacy, the Vernonia Court turned to “the character of the intrusion . . . complained of.” 515 U. S., at 658. Observing that students produce urine samples in a bathroom stall with a coach or teacher outside, Vernonia typed the privacy interests compromised by the process of obtaining samples “negligible.” Ibid. As to the required pretest disclosure of prescription medications taken, the Court assumed that “the School District would have permitted [a student] to provide the requested information in a confidential manner — for example, in a sealed envelope delivered to the testing lab.” Id., at 660. On that assumption, the Court concluded that Vernonia’s athletes faced no significant invasion of privacy.
In this case, however, Lindsay Earls and her parents allege that the School District handled personal information collected under the policy carelessly, with little regard for its confidentiality. Information about students’ prescription drug use, they assert, was routinely viewed by Lindsay’s choir teacher, who left files containing the information unlocked and unsealed, where others, including students, could see them; and test results were given out to all activity sponsors whether or not they had a clear “need to know.” See *849Brief for Respondents 6, 24; App. 105-106, 131. But see id., at 199 (policy requires that “[t]he medication list shall be submitted to the lab in a sealed and confidential envelope and shall not be viewed by district employees”).
In granting summary judgment to the School District, the District Court observed that the District’s “[pjolicy expressly provides for confidentiality of test results, and the Court must assume that the confidentiality provisions will be honored.” 115 F. Supp. 2d 1281, 1293 (WD Okla. 2000). The assumption is unwarranted. Unlike Vernonia, where the District Court held a bench trial before ruling in the School District’s favor, this case was decided by the District Court on summary judgment. At that stage, doubtful matters should not have been resolved in favor of the judgment seeker. See United States v. Diebold, Inc., 369 U. S. 654, 655 (1962) (per curiam) (“On summary judgment the inferences to be drawn from the underlying facts contained in [affidavits, attached exhibits, and depositions] must be viewed in the light most favorable to the party opposing the motion.”); see also 10A C. Wright, A. Miller, & M. Kane, Federal Practice and Procedure § 2716, pp. 274-277 (3d ed. 1998).
Finally, the “nature and immediacy of the governmental concern,” Vernonia, 515 U. S., at 660, faced by the Vernonia School District dwarfed that confronting Tecumseh administrators. Vernonia initiated its drug testing policy in response to an alarming situation: “[A] large segment of the student body, particularly those involved in interscholastic athletics, was in a state of rebellion ... fueled by alcohol and drug abuse as well as the student[s’] misperceptions about the drug culture.” Id., at 649 (internal quotation marks omitted). Tecumseh, by contrast, repeatedly reported to the Federal Government during the period leading up to the adoption of the policy that “types of drugs [other than alcohol and tobacco] including controlled dangerous substances, are present [in the schools] but have not identified themselves as major problems at this time.” 1998-1999 Tecum*850seh School’s Application for Funds under the Safe and Drug-Free Schools and Communities Program, reprinted at App. 191; accord, 1996-1997 Application, reprinted at App. 186; 1995-1996 Application, reprinted at App. 180.2 As the Tenth Circuit observed, “without a demonstrated drug abuse problem among the group being tested, the efficacy of the District’s solution to its perceived problem is . . . greatly diminished.” 242 F. 3d, at 1277.
The School District cites Treasury Employees v. Von Raab, 489 U. S. 656, 673-674 (1989), in which this Court permitted random drug testing of customs agents absent “any perceived drug problem among Customs employees,” given that “drug abuse is one of the most serious problems confronting our society today.” See also Skinner v. Railway Labor Executives’ Assn., 489 U. S. 602, 607, and n. 1 (1989) (upholding random drug and alcohol testing of railway employees based upon industry-wide, rather than railway-specific, evidence of drug and alcohol problems). The tests in Von Raab and Railway Labor Executives, however, were installed to avoid enormous risks to the lives and limbs of others, not dominantly in response to the health risks to users invariably present in any case of drug use. See Von Raab, 489 U. S., at 674 (drug use by customs agents involved in drug interdiction creates “extraordinary safety and national security hazards”); Railway Labor Executives, 489 U. S., at 628 (railway operators “discharge duties fraught with such risks of injury to others that even a momentary lapse of attention can have disastrous consequences”); see *851also Chandler v. Miller, 520 U. S. 305, 321 (1997) (“Von Raab must be read in its unique context”).
Not only did the Vernonia and Tecumseh districts confront drug problems of distinctly different magnitudes, they also chose different solutions: Vernonia limited its policy to athletes; Tecumseh indiscriminately subjected to testing all participants in competitive extracurricular activities. Urging that “the safety interest furthered by drug testing is undoubtedly substantial for all children, athletes and nonath-letes alike,” ante, at 836, the Court cuts out an element essential to the Vernonia judgment. Citing medical literature on the effects of combining illicit drug use with physical exertion, the Vernonia Court emphasized that “the particular drugs screened by [Vernonia’s] Policy have been demonstrated to pose substantial physical risks to athletes.” 515 U. S., at 662; see also id., at 666 (Ginsburg, J., concurring) (Vernonia limited to “those seeking to engage with others in team sports”). We have since confirmed that these special risks were necessary to our decision in Vernonia. See Chandler, 520 U. S., at 317 (Vernonia “emphasized the importance of deterring drug use by schoolchildren and the risk of injury a drug-using student athlete cast on himself and those engaged with him on the playing field”); see also Ferguson v. Charleston, 532 U. S. 67, 87 (2001) (Kennedy, J., concurring) (Vernonia’s policy had goal of “ ‘[djeterring drug use by our Nation’s schoolchildren,’ and particularly by student-athletes, because ‘the risk of immediate physical harm to the drug user or those with whom he is playing his sport is particularly high’ ”) (quoting Vernonia, 515 U. S., at 661-662).
At the margins, of course, no policy of random drug testing is perfectly tailored to the harms it seeks to address. The School District cites the dangers faced by members of the band, who must “perform extremely precise routines with heavy equipment and instruments in close proximity to other students,” and by Future Farmers of America, who *852“are required to individually control and restrain animals as large as 1500 pounds.” Brief for Petitioners 43. For its part, the United States acknowledges that “the linebacker faces a greater risk of serious injury if he takes the field under the influence of drugs than the drummer in the halftime band,” but parries that “the risk of injury to a student who is under the influence of drugs while playing golf, cross country, or volleyball (sports covered by the policy in Ver-nonia) is scarcely any greater than the risk of injury to a student... handling a 1500-pound steer (as [Future Farmers of America] members do) or working with cutlery or other sharp instruments (as [Future Homemakers of America] members do).” Brief for United States as Amicus Curiae 18. One can demur to the Government’s view of the risks drug use poses to golfers, cf. PGA TOUR, Inc. v. Martin, 532 U. S. 661, 687 (2001) (“golf is a low intensity activity”), for golfers were surely as marginal among the linebackers, sprinters, and basketball players targeted for testing in Ver-nonia as steer-handlers are among the choristers, musicians, and academic-team members subject to urinalysis in Tecumseh.3 Notwithstanding nightmarish images of out-of-control flatware, livestock run amok, and colliding tubas disturbing the peace and quiet of Tecumseh, the great majority of students the School District seeks to test in truth are engaged in activities that are not safety sensitive to an unusual degree. There is a difference between imperfect tailoring and no tailoring at all.
The Vernonia district, in sum, had two good reasons for testing athletes: Sports team members faced special health risks and they “were the leaders of the drug culture.” Vernonia, 515 U. S., at 649. No similar reason, and no other tenable justification, explains Teeumseh’s decision to target *853for testing all participants in every competitive extracurricular activity. See Chandler, 520 U. S., at 319 (drug testing candidates for office held incompatible with Fourth Amendment because program was “not well designed to identify candidates who violate antidrug laws”).
Nationwide, students who participate in extracurricular activities are significantly less likely to develop substance abuse problems than are their less-involved peers. See, e. g., N. Zill, C. Nord, & L. Loomis, Adolescent Time Use, Risky Behavior, and Outcomes 52 (1995) (tenth graders “who reported spending no time in school-sponsored activities were ... 49 percent more likely to have used drugs” than those who spent 1-4 hours per week in such activities). Even if students might be deterred from drug use in order to preserve their extracurricular eligibility, it is at least as likely that other students might forgo their extracurricular involvement in order to avoid detection of their drug use. Tecumseh’s policy thus falls short doubly if deterrence is its aim: It invades the privacy of students who need deterrence least, and risks steering students at greatest risk for substance abuse away from extracurricular involvement that potentially may palliate drug problems.4
To summarize, this case resembles Vernonia only in that the School Districts in both cases conditioned engagement in activities outside the obligatory curriculum on random subjection to urinalysis. The defining characteristics of the two programs, however, are entirely dissimilar. The Vernonia district sought to test a subpopulation of students distinguished by their reduced expectation of privacy, their special *854susceptibility to drug-related injury, and their heavy involvement with drug use. The Tecumseh district seeks to test a much larger population associated with none of these factors. It does so, moreover, without carefully safeguarding student confidentiality and without regard to the program’s untoward effects. A program so sweeping is not sheltered by Vemonia; its unreasonable reach renders it impermissible under the Fourth Amendment.
II
In Chandler, this Court inspected “Georgias requirement that candidates for state office pass a drug test”; we held that the requirement “d[id] not fit within the closely guarded category of constitutionally permissible suspicion-less searches.” 520 U. S., at 309. Georgia’s testing prescription, the record showed, responded to no “concrete danger,” id., at 319, was supported by no evidence of a particular problem, and targeted a group not involved in “high-risk, safety-sensitive tasks,” id., at 321-322. We concluded:
“What is left, after close review of Georgia’s scheme, is the image the State seeks to project. By requiring candidates for public office to submit to drug testing, Georgia displays its commitment to the struggle against drug abuse.... The need revealed, in short, is symbolic, not ‘special,’ as that term draws meaning from our case law.” Ibid.
Close review of Tecumseh’s policy compels a similar conclusion. That policy was not shown to advance the “‘special needs’ [existing] in the public school context [to maintain]... swift and informal disciplinary procedures ... [and] order in the schools,” Vernonia, 515 U. S.,.at 653 (internal quotation marks omitted). See supra, at 846-848, 849-853. What is left is the School District’s undoubted purpose to heighten awareness of its abhorrence of, and strong stand against, drug abuse. But the desire to augment eommunica*855tion of this message does not trump the right of persons— even of children within the schoolhouse gate — to be “secure in their persons . . . against unreasonable searches and seizures.” U. S. Const., Arndt. 4.
In Chandler, the Court referred to a pathmarking dissenting opinion in which “Justice Brandéis recognized the importance of teaching by example: ‘Our Government is the potent, the omnipresent teacher. For good or for ill, it teaches the whole people by its example.’ ” 520 U. S., at 322 (quoting Olmstead v. United States, 277 U. S. 438, 485 (1928)). That wisdom should guide decisionmakers in the instant case: The government is nowhere more a teacher than when it runs a public school.
It is a sad irony that the petitioning School District seeks to justify its edict here by trumpeting “the schools’ custodial and tutelary responsibility for children.” Vernonia, 515 U. S., at 656. In regulating an athletic program or endeavoring to combat an exploding drug epidemic, a school’s custodial obligations may permit searches that would otherwise unacceptably abridge students’ rights. When custodial duties are not ascendant, however, schools’ tutelary obligations to their students require them to “teach by example” by avoiding symbolic measures that diminish constitutional protections. “That [schools] are educating the young for citizenship is reason for scrupulous protection of Constitutional freedoms of the individual, if we are not to strangle the free mind at its source and teach youth to discount important principles of our government as mere platitudes.” West Virginia Bd. of Ed. v. Barnette, 319 U. S. 624, 637 (1943).
* * *
For the reasons stated, I would affirm the judgment of the Tenth Circuit declaring the testing policy at issue unconstitutional.
According to Tecumseh’s choir teacher, choir participants who chose not to wear their choir uniforms to school on the days of competitions could change either in “a rest room in a building” or on the bus, where “[m]any of them have figured out how to [change] without having [anyone] ... see anything.” 2 Appellants’ App. in No. 00-6128 (CA10), p. 296.
The Court finds it sufficient that there be evidence of some drug use in Tecumseh’s schools: “As we cannot articulate a threshold level of drug use that would suffice to justify a drug testing program for schoolchildren, we refuse to fashion what would in effect be a constitutional quantum of drug use necessary to show a ‘drug problem.’ ” Ante, at 836. One need not establish a bright-line “constitutional quantum of drug use” to recognize the relevance of the superintendent’s reports characterizing drug use among Tecumseh’s students as “not ... [a] major problefm],” App. 180, 186,191.
Cross-country runners and volleyball players, by contrast, engage in substantial physical exertion. See Vernonia School Dist. 47J v. Acton, 515 U. S. 646, 663 (1995) (describing special dangers of combining drug use with athletics generally).
The Court notes that programs of individualized suspicion, unlike those using random testing, “might unfairly target members of unpopular groups.” Ante, at 837; see also ante, at 841-842 (Breyer, J., concurring). Assuming, arguendo, that this is so, the School District here has not exchanged individualized suspicion for random testing. It has installed random testing in addition to, rather than in lieu of, testing “at any time when there is reasonable suspicion.” App. 197.