This challenge to a federal employee drug testing program raises Fourth Amendment issues both familiar and new to this court. Appellant, the National Treasury Employees Union (NTEU), appeals an order of the district court allowing the United States Department of Agriculture (USDA) to proceed with random urinalysis drug testing of certain USDA motor vehicle operators and "reasonable suspicion" drug testing of other Department workers. See National Treasury Employees Union v. Yeutter, 733 F. Supp. 403 (D.D.C.1990). We find USDA's plan to test the motor vehicle operators at issue here materially indistinguishable from testing previously upheld by this court, and accordingly affirm on that issue. We are unable, however, to approve the Department's reasonable suspicion testing program in its current form. Because USDA regulations authorize urinalysis testing of ordinary employees who are not suspected of on-duty drug use or impairment, and because those regulations mandate needlessly intrusive testing procedures, we find them unconstitutional. We therefore reverse the decision of the district court in part and remand.

Executive Order 12,564, issued September 15, 1986, barred drug use by federal employees on- and off-duty, and directed each executive agency to develop a plan for achieving a "drug-free workplace." Exec. Order No. 12,564, 3 C.F.R. 224 (1987) reprinted in 5 U.S.C. Sec. 7301 note at 909-11 (1988). The Order specified that these plans should provide for voluntary drug testing, testing when "there is a reasonable suspicion that any employee uses illegal drugs," post-accident and unsafe practice testing, testing in connection with drug counseling and rehabilitation, and testing of job applicants. Id. at 910. Responding to this Order, USDA issued its Drug-Free Workplace Program (the "Program") in August 1988. U.S. Dep't of Agric., Drug-Free Workplace Program (Department Personnel Manual Supp. 792-3, issued August 8, 1988).

NTEU challenged several aspects of the Program in its capacity as the collective bargaining representative of more than 800 employees of USDA's Food and Nutrition Service^S, and the district court consolidated that suit with a related action brought by the National Association of Agricultural Employees, representing employees of USDA's Animal and Plant Health Inspection Service (APHIS). Together, the unions sought injunctions against random testing, reasonable suspicion testing, applicant testing, and post-accident testing.

On cross-motions for summary judgment, the trial court permanently enjoined random urinalysis testing of APHIS inspectors and FNS computer specialists, finding that the government lacks a constitutionally sufficient interest in testing these workers. NTEU v. Yeutter, 733 F. Supp. at 409-13. The court refused, however, to enjoin reasonable suspicion testing, post-accident testing, or random testing of FNS motor vehicle operators, concluding in each case that the government has an interest in testing strong enough to override legitimate employee privacy concerns. Id. at 413-17. Finally, the district court rejected the unions' constitutional challenge to applicant testing on standing grounds and dismissed statutory claims under the Civil Service Reform Act of 1978 and the Rehabilitation Act of 1973 as premature. Id. at 418.

On appeal, NTEU challenges those portions of the district court's order allowing USDA to go forward with random urinalysis drug testing of FNS motor vehicle operators and reasonable suspicion testing of all FNS workers. The National Association of Agricultural Employees is not a party to this appeal, and no other provisions of the USDA Program are at issue. Appellant does not challenge the district court's disposition of its statutory claims.

The judicial slate on Fourth Amendment challenges to government drug testing programs is already well-inscribed, with the Supreme Court's pronouncements in Skinner v. Railway Labor Executives Association, 489 U.S. 602, 109 S. Ct. 1402, 103 L. Ed. 2d 639 (1989), and National Treasury Employees Union v. Von Raab, 489 U.S. 656, 109 S. Ct. 1384, 103 L. Ed. 2d 685 (1989), etched most deeply. In Skinner, the Court upheld Federal Railroad Administration regulations that required post-accident drug testing of railroad employees and permitted testing after rules violations or minor accidents, and upon suspicion of on-duty impairment. It stressed the contextual reasonableness of dispensing with warrant and probable cause requirements, 109 S. Ct. at 1415-16, the diminished privacy expectations of railroad workers with respect to inquiries into their health and fitness,, id., 109 S.Ct. at 1418-19, the risk that drug impaired railroad workers could "cause great human loss before any signs of impairment become noticeable to supervisors or others," id., 109 S.Ct. at 1419, and the difficulty of detecting and deterring on-duty drug use through less intrusive means, id., 109 S.Ct. at 1419-20.

In Von Raab, the Court upheld those parts of a Customs Service testing plan that mandated urinalysis of employees directly involved in drug interdiction and those required to carry firearms. The Court followed the analytic path taken in Skinner, weighing the public interest in drug testing against the reasonable privacy expectations of tested employees. 109 S. Ct. at 1392-96. The record did not allow assessment of the reasonableness of testing Customs Service workers with access to classified material, however, so the Court remanded for further proceedings on this issue. Id., 109 S.Ct. at 1396-97.

Skinner and Von Raab establish that compelled urinalysis is a Fourth Amendment search, but also that individualized suspicion of drug use is not an invariable constitutional necessity. Skinner, 109 S. Ct. at 1412-13, 1417; Von Raab, 109 S. Ct. at 1390. Where urinalysis "serves special government needs, beyond the normal need for law enforcement, it is necessary to balance the individual's privacy expectations against the Government's interests to determine whether it is impractical to require a warrant or some level of individualized suspicion in a particular context." Von Raab, 109 S. Ct. at 1390. With this framework in mind, we turn to the challenges appellant brings before us.

We do not start from scratch when evaluating NTEU's challenge to random testing of FNS drivers, a fact that greatly simplifies our balancing task. Because this part of USDA's Program parallels provisions of a Department of Transportation (DOT) drug testing program upheld in American Federation of Government Employees (AFGE) v. Skinner, 885 F.2d 884 (D.C.Cir.1989), cert. denied, --- U.S. ----, 110 S. Ct. 1960, 109 L. Ed. 2d 321 (1990), we need only decide whether the few distinctions between these situations require us to reach a different result here.

[S]trong safety interests support the testing of most Department [of Transportation] motor vehicle operators, who are responsible for, inter alia, the transportation of visiting foreign dignitaries and key Department officials and the operation of passenger-laden shuttle buses. Shuttle buses transport as many as 1,200 passengers each day. Thus, obvious safety interests support the testing of the majority of the Department's motor vehicle operators.

885 F.2d at 892 (citations omitted). Safety interests alone were not sufficient to validate random testing of DOT mail van operators, but the court found that strong national security concerns supported testing of these employees. Id. at 892-93.

The procedures for random testing of FNS drivers are identical in relevant respects to those at issue in AFGE v. Skinner, as both USDA and DOT follow drug testing regulations issued by the Department of Health and Human Services. See AFGE v. Skinner, 885 F.2d at 887; U.S. Dep't of Agric., Drug-Free Workplace Program Sec. 13-1(A); see also Mandatory Guidelines for Federal Workplace Drug Testing Programs, 53 Fed.Reg. 11,970 (1988). Thus, only the specific duties of FNS and DOT drivers and differences in reasonable privacy expectations between the two groups might justify different outcomes.

NTEU stresses that whereas DOT motor vehicle operators routinely drove shuttle-buses, FNS drivers primarily chauffeur officials and deliver documents, with shuttle responsibilities limited to filling in for a private bus driver during lunch hours and sick days. From this the union concludes that the government's safety interest is so much weaker here than in AFGE v. Skinner that the constitutional balance tips against testing.

But in AFGE v. Skinner the court considered DOT's safety interest in testing its drivers "obvious," 885 F.2d at 892, a characterization that should discourage line-drawing based upon the number of passengers carried in an average day. And in Jones v. McKenzie, 833 F.2d 335 (D.C.Cir.1987), vacated sub nom. Jenkins v. Jones, 490 U.S. 1001, 109 S. Ct. 1633, 104 L. Ed. 2d 149, modified, 878 F.2d 1476 (D.C.Cir.1989), this court upheld urinalysis drug testing of school bus attendants as well as drivers, refusing to draw a relevant distinction between the two groups:

While the safety concern may be somewhat greater for a school bus driver, it is still quite significant in the case of an employee who is responsible for supervising, attending and carrying handicapped children. For example, the danger to a young, handicapped child, should she be dropped by an attendant or ignored while crossing the street, is obvious.

Id. at 340. If the government has a "compelling" interest in testing school bus attendants for drug use, as this court confirmed in Jones upon remand from the Supreme Court, see Jones v. Jenkins, 878 F.2d at 1477, it certainly has a similarly strong interest in testing FNS drivers who are responsible for the safety of their passengers--however many in number.

Nor do differences in the privacy expectations of FNS and DOT motor vehicle operators suggest a different outcome. It is true that the drivers in AFGE v. Skinner, unlike FNS drivers, underwent background investigations that should have diminished their expectations of privacy with respect to testing for use of illegal drugs. AFGE v. Skinner, 885 F.2d at 893. But the court did not rely on that fact in approving testing of those DOT drivers who carried passengers, implicitly suggesting the gravity of the safety interests involved. See id. at 892-93. Moreover, FNS drivers should be aware that their physical and mental alertness is of concern to USDA: Prior to employment they must submit a doctor's Certificate of Medical Examination warranting their mental and physical suitability for the job, and every four years they must certify on a Physical Fitness Inquiry form that they do not have a drug habit. We conclude that the reasonable privacy expectations of FNS drivers are not so different from those of DOT drivers as to require departure from the result reached in AFGE v. Skinner.

NTEU claims that upholding testing of FNS drivers will open the door to testing of all government employees who drive to work, or even all licensed drivers. This is more rhetoric than argument. It is not obvious to us that the government could show a special need, unrelated to law enforcement, to test drivers who do not carry passengers; nor do most drivers have diminished privacy expectations with respect to drug testing. True, the Supreme Court "has quite clearly eschewed an approach to drug testing based on bright lines," Harmon v. Thornburgh, 878 F.2d 484, 490 n. 9 (D.C.Cir.1989), cert. denied, --- U.S. ----, 110 S. Ct. 865, 107 L. Ed. 2d 949 (1990), but surely such a line could be drawn between testing individuals whom the government pays to transport passengers safely and testing individuals who drive on their own time.

All USDA employees are subject to reasonable suspicion urinalysis under the Program. If an FNS employee is suspected of on-duty or off-duty drug use, his or her supervisor will inform the Service's Personnel Director, who makes the final determination of whether or not reasonable suspicion exists and, if reasonable suspicion is found, compiles a written record of the evidence suggesting drug use. Reasonable suspicion may be based on, "among other things":

1. Observable phenomena, such as direct observation of drug use or possession and/or the physical symptom(s) of being under the influence of a drug.

2. A pattern of abnormal conduct or erratic behavior.

3. Arrest or conviction for a drug-related offense, or the identification of an employee as the focus of a criminal investigation into illegal drug possession, use, or trafficking.

4. Information provided either by reliable and credible sources or independently corroborated.

5. Newly discovered evidence that the employee has tampered with a previous drug test.

U.S. Dep't of Agric., Drug-Free Workplace Program Sec. 10-1(A). "Although reasonable suspicion testing does not require certainty," USDA regulations specify, " 'mere hunches' are not sufficient to meet this standard." Id.

Urine samples are analyzed for the presence of marijuana, cocaine, opiates, amphetamines, PCP, or, with the approval of the Department of Health and Human Services, for other controlled substances. Id. at Appendix B-1. Subject to restrictions imposed by the Civil Service Reform Act of 1978, employees who test positive may suffer "the full range of disciplinary actions, including removal." Id. at Sec. 8-1(C).

The immunoassay and gas chromatography/mass spectrometry testing techniques required by federal drug testing guidelines are highly accurate if properly employed. Skinner, 109 S. Ct. at 1409 n. 3; see Mandatory Guidelines for Federal Workplace Drug Testing Programs, 53 Fed.Reg. at 11,983. However, "[a] single positive urine test result is silent as to when or how much of the drug was taken, the pattern of the employee's drug use, or whether the employee was intoxicated when the test was given." National Fed'n of Fed. Employees (NFFE) v. Cheney, 884 F.2d 603, 609 (D.C.Cir.1989), cert. denied, --- U.S. ----, 110 S. Ct. 864, 107 L. Ed. 2d 948 (1990); accord Skinner, 109 S. Ct. at 1409 (blood is the only body fluid that can indicate current drug or alcohol impairment); AFGE v. Skinner, 885 F.2d at 896 (urinalysis testing fails to differentiate on- and off-duty drug impairment). In Von Raab, the Supreme Court noted evidence that "the time it takes for particular drugs to become undetectable in urine can vary widely depending on the individual, and may extend for as long as 22 days." 109 S. Ct. at 1396. Thus, the fact that an employee tests positive does not necessarily indicate on-the-job impairment, very recent drug use, or habitual drug use.

The permissibility of a government search is determined " 'by balancing its intrusion on the individual's Fourth Amendment interests against its promotion of legitimate government interests.' " Skinner, 109 S. Ct. at 1414 (quoting Delaware v. Prouse, 440 U.S. 648, 654, 99 S. Ct. 1391, 1396, 59 L. Ed. 2d 660 (1979)). Since urinalysis drug testing of federal employees implicates Fourth Amendment privacy interests, Von Raab, 109 S. Ct. at 1390, and since the government disclaims any law enforcement purpose underlying the Drug-Free Workplace Program, we may uphold the Program only if other legitimate government interests supporting the program outweigh employee privacy interests.

In O'Connor v. Ortega, 480 U.S. 709, 107 S. Ct. 1492, 94 L. Ed. 2d 714 (1987), the Supreme Court suggested that warrantless searches of ordinary government employees and their possessions can be justified, if at all, only when a nexus to work responsibilities is shown: "In the case of searches conducted by a public employer, we must balance the invasion of the employees' legitimate expectations of privacy against the government's need for supervision, control, and the efficient operation of the workplace." Id. at 719-20, 107 S.Ct. at 1499 (plurality opinion) (approving warrantless search of doctor's office by state hospital officials). Justice Scalia, concurring in the judgment, also distinguished between work-related and other government interests, arguing that "government searches to retrieve work-related materials or to investigate violations of workplace rules--searches of the sort that are regarded as reasonable and normal in the private-employer context--do not violate the Fourth Amendment." Id. at 732, 107 S.Ct. at 1505.

Von Raab and Skinner accord with this understanding. Von Raab recognized a government interest in off-duty drug use only with respect to those Customs Service employees with exceptional duties (drug in terdiction,, use of firearms, and access to "truly sensitive information"). See 109 S. Ct. at 1393, 1396. Similarly, Skinner 's approval of drug testing turned not on the illegality of drug use, but on the "separate and far more dangerous wrong" of performing hazardous tasks while drug- or alcohol-impaired. See 109 S. Ct. at 1421. Other than the motor vehicle operators previously discussed, we know of no FNS employees whose jobs bring them within these safety, security, and drug interdiction exceptions.

Decisions of this court also suggest that, outside of the law enforcement context, the government's legitimate interest in employee drug testing extends no further than its interest in workplace conduct and performance of work responsibilities. In Harmon v. Thornburgh, we stated in the course of reviewing a Department of Justice random urinalysis plan that "federal employment alone is not a sufficient predicate for mandatory urinalysis," and that "the government may search its employees only when a clear, direct nexus exists between the nature of the employee's duty and the nature of the feared violation." 878 F.2d at 490. Similarly, in NFFE v. Cheney we refused to allow random testing of Army personnel responsible for handling and analyzing urine samples "absent a causal connection between the employees' duties and the feared harm." 884 F.2d at 614.

The government counters that it has a legitimate interest in deterring drug use that might affect work performance, that employees who use drugs off the job risk performance-impairing addiction, that off-duty drug users may buy drugs at work or steal to support their habit, and that drug use by federal workers may erode public confidence in the employing agency. But the government has not produced evidence that might establish a relationship between off-the-job drug use and job performance; its arguments rely either on an unsupported connection between off-duty drug use and government efficiency or on general integrity and deterrence rationales of the sort rejected in Harmon and Cheney. The government has ample latitude to intervene before its fears about off-duty drug use are realized in the form of poor work performance, as law enforcement officials may investigate the worker under appropriate constitutional and statutory restrictions. Similarly, the government may investigate and punish wrongdoing within the walls of its offices as an employer and as a law enforcer, again under applicable legal constraints. What the government may not do in the case of ordinary employees is justify drug testing procedures that intrude upon constitutionally protected privacy interests with speculation about possible future job impairment or rules violations.

We hold, therefore, that the USDA Program is unconstitutional insofar as it authorizes mandatory drug testing of FNS workers who do not hold safety- or security-sensitive jobs, absent reasonable suspicion of on-duty drug use or drug-impaired work performance.

When, however, reasonable suspicion of drug use arises from evidence of on-the-job drug use or impairment, USDA's interest in efficient operations may justify departure from warrant and probable cause requirements. See Skinner, 109 S. Ct. at 1414. As the Supreme Court noted in O'Connor v. Ortega, the delay in correcting employee misconduct that would result from adherence to warrant and probable cause requirements may hinder the government's work, and it would in any event be unrealistic to expect all agency supervisors to be schooled in nuances of the probable cause standard. See 480 U.S. at 724-25, 107 S.Ct. at 1501-02; see also Von Raab, 109 S. Ct. at 1391 ("[R]equiring the Government to procure a warrant for every work-related intrusion would conflict with the common sense realization that government offices could not function if every employment decision became a constitutional matter." (internal quotations omitted)).

We are mindful that mandatory urinalysis represents a substantial intrusion on employee privacy, and that most FNS workers have less reason to expect inquiries into their physical and mental fitness than Customs Service employees or railroad engineers. So too, we reject the government's abortive suggestion that FNS' mission to advance human nutrition and protect the agricultural economy bears on the outcome of this case. Still, we cannot conclude that employee privacy concerns outweigh the government's interest in testing employees reasonably suspected of using drugs, or of being drug-impaired, while on duty.

The Supreme Court has made it clear that the very fact of individualized suspicion goes far toward making a search reasonable where the government has a legitimate interest in confirming the alleged violation. See Skinner, 109 S. Ct. at 1417 (individualized suspicion usually necessary when probable cause requirement waived); Ortega, 480 U.S. at 726, 107 S.Ct. at 1502 ("Ordinarily, a search of an employee's office by a supervisor will be 'justified at its inception' when there are reasonable grounds for suspecting that the search will turn up evidence that the employee is guilty of work-related misconduct...."). The fact that urinalysis detects drug use in an employee's private life is a significant concern, but not one so strong as to negate the link between positive test results and on-duty misconduct. A positive drug test would tend to confirm the information that gave rise to testing and might also "provide the basis for further investigative work designed to determine whether the employee used drugs at the relevant times." Skinner, 109 S. Ct. at 1421.

We therefore reject NTEU's core challenge to reasonable suspicion testing where on-duty drug use or impairment is suspected.

Alternatively, NTEU argues that all reasonable suspicion testing of FNS employees is constitutionally defective because FNS supervisors have not been given sufficient training or guidelines for detecting drug impairment. We agree that the reasonableness of a suspicion of drug use formed solely on the basis of "abnormal conduct" is not self-evident, especially since the government's own expert suggests that drug intoxication may produce "few overt signs." Declaration of Robert L. DuPont, Jr., M.D. at 2, reproduced in Joint Appendix at 242, 243. So too, we recognize that plausible objections to the adequacy of USDA's video training program might be raised by a tested employee challenging a specific finding of reasonable suspicion. But none of this is fatal in the context of a facial challenge. Since this court cannot say that supervisors necessarily exercise "unbridled discretion" when they follow USDA regulations, NTEU's arguments must await consideration in a challenge to the regulations as applied. See Skinner, 109 S. Ct. at 1416 n. 6, 1421 n. 10; Schaill v. Tippecanoe County School Corp., 864 F.2d 1309, 1322 n. 19 (7th Cir.1988) (assuming good faith compliance with drug testing guidelines for purposes of facial challenge).

NTEU also challenges the Program's procedural guidelines to the extent that they require visual observation of all employees who are ordered to undergo reasonable suspicion urinalysis. Whereas most tested employees and job applicants are allowed to provide their urine sample without direct visual observation unless testing personnel have specific reason to believe that they may try to cheat on the test, USDA assumes in the case of any employee called in for reasonable suspicion testing that the individual "may alter or substitute the specimen to be provided." Accordingly, a collection monitor of the same sex will watch the employee urinate. U.S. Dep't of Agric., Drug-Free Workplace Program Sec. 13-1(B). This provision is distinct and clearly severable from those that govern reasonable suspicion testing generally, so it is appropriate to measure the observation requirement itself against the core constitutional test of reasonableness. Because we can discern no weighty government interest in observation that counterbalances its intrusion on employee privacy, we hold that this procedural provision violates the Fourth Amendment.

Neither Von Raab nor Skinner directly addressed visual monitoring procedures, but both decisions contain language relevant to our inquiry. In Von Raab the Court held that the absence of visual observation "significantly" diminishes the intrusion on employee privacy, 109 S. Ct. at 1396 n. 4, suggesting that USDA's observation requirement should be justified by strong evidence of necessity. On the other hand, the Court noted in Skinner that the reasonableness of a given procedure does not necessarily depend upon the availability of less intrusive means, that visual monitoring contributes to the integrity of testing as a general matter, and that courts should avoid "second-guessing" reasonable conclusions drawn by an agency about the necessity of a given testing program. 109 S. Ct. at 1418, 1419 n. 9. See also NFFE v. Cheney, 884 F.2d at 611 (Army better positioned than court to assess the efficacy of urinalysis testing).

We need not resolve whatever tensions may exist in Supreme Court precedent to determine that the visual monitoring regulation now at issue does not significantly improve testing accuracy. In Von Raab, the government strongly defended the accuracy of urinalysis testing. This led the Court to conclude that, even absent routine visual monitoring, "no employee reasonably can ... expect attempts at adulteration to succeed, in view of the precautions taken by the sample collector to ensure the integrity of the sample." 109 S. Ct. at 1396. (These precautions include coloring toilet water to discourage adulteration of the urine specimen and taking away personal belongings and unnecessary outer garments that might conceal substances intended to foil the test. Mandatory Guidelines for Federal Workplace Drug Testing Programs, 53 Fed.Reg. at 11,980.) Here, the government stresses the vulnerability of drug tests to cheating, but NTEU reminds the government--and this court--that the government's success in Von Raab negates its current expressions of concern.

Since nothing in the USDA regulations requires administrators to give employees who are suspected of drug use prior notice of testing, these employees will have greater difficulty arming themselves to foil a drug test than, for example, USDA job applicants, who have prior warning of drug testing but are not watched while urinating. On average, employees suspected of drug use may be more likely to cheat on a urine test than randomly tested employees or job applicants; it does not follow, though, that drug users are more likely to cheat on surprise reasonable suspicion drug tests than on previously announced random tests or applicant tests. We also have heard no reason why the standard monitoring procedures--collecting outer-garments, dying toilet water, listening for urination--are inadequate to keep suspected drug users from cheating if they are adequate to prevent unsuspected drug users from doing the same.

Finally, we note that the agency responsible for developing government drug testing guidelines, the Department of Health and Human Services, has concluded that visual observation is unnecessary to ensure accuracy except where there is specific "reason to believe that a particular individual may alter or substitute the specimen to be provided." Mandatory Guidelines for Federal Workplace Drug Testing Programs, 53 Fed.Reg. at 11,981. This determination by a federal agency with relevant expertise makes us especially hesitant to accept the government's say-so that automatic visual monitoring significantly improves testing accuracy.

Accordingly, we invalidate the Program's guidelines for visual monitoring to the extent that they require observation whenever reasonable suspicion testing is conducted, regardless of the circumstances. Appellant has not challenged, and we do not address, other provisions that authorize individualized determinations of a need for observation. It is implicit in our holding, however, that indications of drug use sufficient to warrant reasonable suspicion testing will not necessarily establish a legitimate need to observe urination in all instances.D. Remedy

We are mindful that injunctions issued by federal courts should be narrowly tailored to remedy the harm shown. Gulf Oil Corp. v. Brock, 778 F.2d 834, 842 (D.C.Cir.1985). At the same time, we face remedial issues similar to those presented in Harmon v. Thornburgh, where the court found it inappropriate to redraw employee categories drawn by the Department of Justice so as to allow random testing of some employees within those categories while prohibiting testing of others. See 878 F.2d at 493-96. Having carefully reviewed the record before us, we conclude that the dispositive concerns in Harmon are not present here: Those portions of the USDA Program invalidated by this decision can be "clearly defined"; we are confident that the Department of Agriculture would wish to carry out its testing plan to the extent that it is constitutionally permissible under our analysis; and issuing a narrowly tailored remedial order does not require us to reach unnecessary or undeveloped constitutional issues. See id. at 493-94. Nor will a focused injunction involve this court in administrative or regulatory decisions for which the judiciary is unsuited.

Accordingly, appellees must be enjoined from testing FNS employees under the Program's reasonable suspicion testing provisions except where: 1) there is reasonable, articulable, and individualized suspicion of on-duty drug use or impairment; or 2) the employee at issue holds a security- or safety-sensitive position as to which mandatory random urinalysis may be conducted under this opinion and the opinion and order of the district court. FNS motor vehicle operators fall within this latter category. Appellees also must be enjoined from requiring visual observation of urination when the need for observation is established by nothing more than the fact of reasonable suspicion testing.

We affirm the district court's opinion and order insofar as it allows USDA to proceed with random urinalysis drug testing of FNS motor vehicle operators, finding no basis to distinguish these employees from those subject to random testing under the DOT plan at issue in AFGE v. Skinner. We also affirm the district court's decision to the extent that it permits reasonable suspicion testing of these motor vehicle operators to go forward.

With regard to reasonable suspicion testing of FNS employees who do not hold security- or safety-sensitive positions, however, we remand to the district court with instructions to modify the injunction so as to prohibit reasonable suspicion urinalysis drug testing that is not supported by reasonable, articulable, and individualized suspicion of on-duty drug use or impairment. We further instruct the district court to enjoin appellees from requiring visual observation of urination whenever an employee is ordered to undergo reasonable suspicion testing, and from authorizing observation of urination during reasonable suspicion drug tests absent an individualized determination that visual monitoring is warranted.

The case is remanded to the district court for further proceedings consistent with this opinion.

So ordered.