JUDGMENT

Plaintiffs sued to enjoin an allegedly unconstitutional drug testing program. The program was instituted by the United States Department of Justice for employees of the Federal Bureau of Prisons (the “Bureau”). Plaintiffs are a union that represents Bureau employees. In June 1988, this Court granted plaintiffs’ motion for a preliminary injunction and enjoined the testing of any Bureau employee absent reasonable suspicion that drug use by that employee impaired his or her ability to perform official duty. American Federation of Government Employees, Council 33 v. Meese, 688 F.Supp. 547, 556 (N.D.Cal.1988). In September 1989, upon reconsideration of the case in light of decisions by the United States Supreme Court in Skinner v. Railway Labor Executives’ Ass’n, 489 U.S. 602 , 109 S.Ct. 1402 , 103 L.Ed.2d 639 (1989) (“Skinner”), and National Treasury Employees Union v. Von Raab, 489 U.S. 656 , 109 S.Ct. 1384 , 103 L.Ed.2d 685 (1989) (“Von Raab”), this Court reaffirmed the preliminary injunction without modification. American Federation of Government Employees, Council 33 v. Thornburgh, 720 F.Supp. 154 (N.D.Cal.1989). Before that reaffirmation, the Court had denied defendants’ motion for summary judgment on plaintiffs’ claims under the Fourth Amendment and the Civil Service Reform Act of 1974, 5 U.S.C. § 2302 (b)(10).

Defendants now renew their motion for summary judgment on these claims. Supporting this renewed motion are a number of decisions by the Ninth Circuit and other jurisdictions upholding drug-testing programs with many of the same components as the program proposed by defendants. Defendants have also modified their original proposal in an attempt to bring it within constitutional bounds under prevailing case law. Plaintiffs agree that summary adjudication is appropriate but they raise objections to various aspects of defendants’ modified proposal.

I. BACKGROUND

The drug-testing program proposed by the Bureau in 1988 called for the random testing of all Bureau employees. AFGE v. Meese, 688 F.Supp. at 549 . The program also provided for the regular testing of job applicants, probationary employees, and management employees; testing in connection with on-the-job accidents or unsafe activities; and testing upon reasonable suspi *1469 cion that an employee was under the influence of or using drugs. Id. at 549 and n. 2. This Court enjoined all proposed testing save that based on reasonable suspicion that drug use by an employee impaired his or her ability to perform official duty. Id. at 556 . Several findings underlay this Court’s decision. First, the Court found that the specific testing procedure — urinalysis — was both highly intrusive and minimally conclusive. Id. at 551-52 . The Court also found the government interests purportedly served by the testing — safety, prevention of corruption, and preservation of public confidence in the Bureau — less than compelling. Id. at 552-54 . The Court specifically noted the inability of urinalysis to demonstrate current impairment and the complete absence of any evidence indicating the existence or imminence of safety problems or employee corruption related to drug use by Bureau employees. Id. at 553-54 . The Court concluded that less intrusive measures could deter the introduction of drugs by employees at Bureau institutions. Id. at 554 .

Defendants’ new program proposes testing in the same general categories initially outlined. The proposed random testing is now narrowed, however, to cover certain “testing-designated positions.” Both the “post-accident” and “reasonable suspicion” components have also been somewhat narrowed. Plaintiffs nevertheless maintain their constitutional challenge to these three components. 1

II. DISCUSSION

A. Standards.

It is undisputed that mandatory urinalysis is a “search” under the Fourth Amendment. Skinner, 489 U.S. at 617 , 109 S.Ct. at 1413. Whether it is an unreasonable search depends upon its nature and all the circumstances surrounding it. Id. at 619 , 109 S.Ct. at 1414 (citation omitted). Neither a warrant nor probable cause is required where “special needs, beyond the normal need for law enforcement make such a prerequisite impracticable.” Id. (citations omitted). The government’s interest in operating a government office can present such “special needs” to justify a departure from the usual Fourth Amendment requirements. Id. at 620, 109 S.Ct. at 1414-15. When faced with such special needs, this Court must “balance the governmental and privacy interests” to determine whether a search is reasonable and thus constitutional. Id. at 619, 109 S.Ct. at 1414; Von Raab, 489 U.S. at 665, 679 , 109 S.Ct. at 1390, 1397-98 . The Ninth Circuit has followed this balancing approach in four recent decisions upholding drug-testing programs that allow the testing of employees absent reasonable suspicion of on-the-job impairment. Railway Labor Executives’ Ass’n v. Skinner, 934 F.2d 1096 (9th Cir.1991) (“RLEA 17”); Int’l Brotherhood of Teamsters v. Dept. of Transportation, 932 F.2d 1292 (9th Cir.1991) {“Teamsters”) Int’l Brotherhood of Electrical Workers, Local 1245 v. Skinner, 913 F.2d 1454 (9th Cir.1990) (“IBEW”); Bluestein v. Skinner, 908 F.2d 451 (9th Cir.1990), cert. denied, — U.S.-, 111 S.Ct. 954 , 112 L.Ed.2d 1042 (1991) (“Bluestein”).

“Special needs” exist here under the meaning of the recent cases. The Bureau plan is not designed to enforce drug laws but to ensure the safe and effective discharge of duties by Bureau personnel, particularly in the context of Bureau prisons. See Federal Bureau of Prisons Drug-Free Workplace Program Statement (“Program Statement”) at 1, 3-4. Test results may not be used in a criminal prosecution without the employee’s consent. Pub.L. 100-71, 5 U.S.C. § 7301 note, at § 3(e); cf. Von Raab, 489 U.S. at 666 , 109 S.Ct. at 1390-91 . The balancing test is therefore appropriate here. See Teamsters, 932 F.2d at 1299 ; Von Raab, 489 U.S. at 666 , 109 S.Ct. at 1390-91 .

B. Random Testing.

1. Privacy interests.

This Court found urinalysis — the Bureau’s proposed testing method — to be *1470 highly intrusive. AFGE v. Meese, 688 F.Supp. at 551 . 2 Since that finding, the Supreme Court has rejected the notion that urinalysis is always a substantial privacy invasion. Von Raab, 489 U.S. at 671 , 109 S.Ct. at 1393 . That Court has also identified a number of factors that may reduce privacy expectations and thus the intrusiveness of the testing. These factors include prior notice, see id. at 672-73 n. 2,109 S.Ct. at 1394 n. 2; limited discretion in choosing the tested employees, see Skinner, 489 U.S. at 634 , 109 S.Ct. at 1422 ; and the particular employment context, see id. at 627-28 , 109 S.Ct. at 1418-19; Von Raab, 489 U.S. at 677 , 109 S.Ct. at 1396-97 . These same factors minimize the privacy expectations of at least certain Bureau employees under the testing program proposed here. The Bureau will issue a specific notice to all employees and a copy of the revised program long before the testing begins. Fourth Quinlan Decl. at 11 2. 3 The random selection of employees to be tested would leave no room for supervisory discretion. IBEW, 913 F.2d at 1460 ; First Quinlan Decl. at II 8(b) and Exh. E. Bureau employees undergo urinalysis testing and background and integrity checks when hired, and whenever they report to work at a Bureau institution they are subject to routine searches of their person and property and, upon reasonable suspicion of drug use or criminal activity, a visual search, pat search, urine surveillance test, breathalyzer test, or “other comparable test” whenever they report to work at an institution. First Quinlan Decl. at 11 5 and Exhs. B & C. It therefore appears that employees who work at Bureau institutions have, while on the job, privacy expectations “markedly less than those of the public in general.” Teamsters, 932 F.2d at 1300 . See also Von Raab, 489 U.S. at 677-78 , 109 S.Ct. at 1396-97 ; Skinner, 489 U.S. at 627-28 , 109 S.Ct. at 1418-19 ; Connelly v. Newman, 753 F.Supp. 293, 296 (N.D.Cal.1990) (privacy expectations of investigators in Office of Personnel Management are comparable to those of railroad employees considered in Skinner since both work “in an industry subject to heavy government regulation”); NFFE v. Cheney, 884 F.2d 603 , 612-613 (D.C.Cir.1989), cert. denied, 493 U.S. 1056 , 110 S.Ct. 864 , 107 L.Ed.2d 948 (1990) (pre-employment screenings and medical tests lower reasonable privacy expectations of Army civilian guards); AFGE v. Cavazos, 721 F.Supp. 1361, 1374-75 (D.D.C.1989), affd in part and vacated in part, 926 F.2d 1215 (1991) (Department of Education positions designated critical and sensitive and for which employees must complete application form inquiring about personal lives and prior drug use have reduced privacy expectations).

The Bureau’s urinalysis procedures follow the mandatory guidelines promulgated by the Department of Health and Human Services (HHS). 53 Fed.Reg. 11970, 11979 ff. (1988); Program Statement at 5. The Supreme Court reviewed and approved these procedures in Von Raab. 489 U.S. at 661 -62 n. 1 and 672-73 n. 2, 109 S.Ct. at 1388-89 n. 1 and 1394 n. 2. That the Bureau’s program includes a random testing component, unlike the programs upheld in Skinner and Von Raab , would not appear to change the privacy calculus significantly. See RLEA II, 934 F.2d at 1098-1100 ; Bluestein, 908 F.2d at 456-457 ; cf. Harmon v. Thornburgh, 878 F.2d 484, 489, 492 (D.C.Cir.1989), cert. denied, 493 U.S. 1056 , 110 S.Ct. 865 , 107 L.Ed.2d 949 (1990). In fact, the Ninth Circuit has found random testing limits the potential intrusiveness by eliminating supervisory discretion and thus the possibility of actual or perceived harassment. See IBEW, 913 F.2d at 1460 ; cf. AFGE v. Skinner, 885 F.2d 884, 891 (D.C.Cir.1989), cert. denied, 495 U.S. 923 , 110 S.Ct. 1960 , 109 L.Ed.2d 321 (1990). It has also approved random testing as particularly well-suited to deterring drug use. RLEA II, 934 F.2d at 1099 ; IBEW, 913 F.2d at 1460 ; cf. AFGE v. Skinner, 885 F.2d at 891 . Finally, the fact that the less intrusive measures may exist — an issue *1471 given weight by this Court, see AFGE v. Meese, 688 F.Supp. at 554 — is not prohibitive. See Skinner, 489 U.S. at 629 n. 9, 109 S.Ct. at 1419 n. 9; cf. NFFE v. Cheney, 884 F.2d at 610. This Court therefore has no choice but to find that random urinalysis of employees working at Bureau institutions implicates diminished privacy interests, given the already substantially reduced privacy expectations of these employees and the binding authority upholding urinalysis testing under methods identical to those proposed here. Now the government’s interests must be balanced against these privacy interests.

2. Government interests.

The Supreme Court has identified three government interests that can support drug testing without particular suspicion of an individual employee’s drug use and impairment. These are (1) promotion of the public safety, (2) protection of truly sensitive information, and (3) maintenance of employee integrity. See Skinner, 489 U.S. at 628 , 109 S.Ct. at 1418-19 ; Von Raab, 489 U.S. at 670-71, 677 , 109 S.Ct. at 1393-94, 1396-97 . Defendants advance precisely these three rationales to justify their proposed program.

a. “Public Safety.”

Under the “public safety” rationale, defendants would test all those employees who may potentially use firearms in the course of their Bureau duties. This group includes those employees in positions designated as “primary law enforcement” positions, which comprises all employees who work at correctional institutions and includes jobs from “correction officer” to “upholstering” and “animal caretaking”; and employees in “secondary law enforcement” positions, which comprise all employees previously in primary law enforcement positions who have now been re-assigned to noninstitutional settings. Secondary law enforcement employees include “corrections officers” as well as non-institutional employees engaged in myriad activities such as “cooking” and “accounting.” Defendants have provided evidence that all employees in both primary and secondary law enforcement positions are trained in the use of firearms and, while none of them regularly carry weapons, any of these employees be issued firearms and called upon to use them in emergency situations. In addition to these employees, defendants would also test physicians and dentists under the “public safety” rubric.

There is no doubt that the government’s interest in testing employees who may carry firearms in the regular course of their duties is compelling. Von Raab, 489 U.S. at 670 , 109 S.Ct. at 1393 . Defendants insist that this interest justifies the testing of all primary and secondary law enforcement Bureau employees because all are trained in the use of firearms and might be called upon to use them in emergency situations. Defendants insist that the frequency of firearm use is insignificant since the magnitude of the harm caused by its misuse is “catastrophic.” Defendants’ authorities for this proposition, however, allow the testing of employees who regularly carry firearms. See NFFE v. Cheney, 884 F.2d at 612 (Army guards); AFGE Local 1533 v. Cheney, 754 F.Supp. 1409, 1423-24 (N.D.Cal.1990), aff'd. on other grounds 944 F.2d 503 (9th Cir.1991) (Navy civilian law enforcement personnel); AFGE v. Cavazos, 721 F.Supp. at 1372 (armed civilian guard); Guiney v. Roache, 873 F.2d 1557, 1558 (1st Cir.), cert. denied, 493 U.S. 963 , 110 S.Ct. 404 , 107 L.Ed.2d 370 (1989) (police officers); but see Seelig v. Koehler, 76 N.Y.2d 87 , 556 N.Y.S.2d 832, 836 , 556 N.E.2d 125, 129 (N.Y.), cert. denied, 111 S.Ct. 134 (1990) (upholding random urinalysis of state jail guards who do not regularly carry firearms). The federal decisions upholding, on “public safety” grounds, the random testing of employees who do not regularly carry firearms do so since the particular occupations involve at all times the discharge of duties “fraught with such risks of injury to others that even a momentary lapse of attention can have disastrous consequences.” Von Raab, 489 U.S. at 670 , 109 S.Ct. at 1393 (Customs officers carrying firearms); Skinner, 489 U.S. at 628 , 109 S.Ct. at 1418-19 (railroad employees involved in train accidents); RLEA II, 934 F.2d at 1097 n. 1 (railroad employees in *1472 “safety-sensitive” positions); Bluestein, 908 F.2d at 456 (passenger airline personnel); IB EW, 913 F.2d at 1462 (pipeline workers); Teamsters, 932 F.2d at 1306 (commercial bus and truck drivers), and cases cited therein. Such great danger may very well exist with Bureau employees who carry firearms on some regular basis, but clearly it does not exist with those who do not. That a vast majority of Bureau employees is trained to use firearms, and that occasional emergencies require the assignment to hazardous duties and issuance of firearms even to clerical workers, does not create the kind of constant danger to justify the random testing of thousands of Bureau employees who do not routinely, and most probably never will, carry firearms. 4

Defendants insist that no Bureau employee “regularly” carries firearms and that their issuance is “dictated by the situation.” For instance, the Bureau assigns employees to institution tower assignments on a rotational basis, and while at that post employees would have access to firearms. Sixth Quinlan Deck at 1112. Similarly, Bureau employees other than correctional officers might be assigned for specific contingencies such as “perimeter patrol” or “escort duty” which require the carrying of firearms. Id. Defendants do note, however, that those employees in primary law enforcement positions are more likely to contend with these contingencies than those in secondary law enforcement positions, and that for such contingencies correctional officers would ordinarily be called upon first. Id. at 1115. The prospect that, say, a secretary might be issued a firearm thus appears remote. Any concomitant danger to public safety as a result of drug impairment by this secretary — or any of thousands of other Bureau employees — is thus greatly attenuated.

Accordingly, the Court finds that only those employees in primary law enforcement positions who, in the regular course of their duties, are issued or given access to firearms for use on a daily or weekly basis may be subject to random testing under this “public safety” rationale. It should be noted that the Court found diminished privacy expectations specifically for those employees who work at Bureau institutions; the category for testing in the interest of public safety is equally circumscribed. The “for use” language simply distinguishes employees who are given access to the firearms for potential use from those who are given such access for transportation, maintenance, inventory, etc.

Defendants also propose to test physicians and dentists on public safety grounds, because of the necessity that medical professionals be scrupulously accurate in every phase of their relationships with patients and the great harm that can attend any misdiagnosis or mistreatment. Plaintiffs respond only that their job duties can not subject these employees to random drug testing. Plaintiffs are incorrect. It requires no detailed knowledge of the medical field to recognize that doctors must attend to each and every one of their patients with the utmost alertness and care. Public safety and, in some cases, life and death may hang in the balance. Doctors must correctly diagnose ailments and either treat patients or supervise others who do. Treatments may include surgery or the use of powerful drugs. Doctors must be able to respond immediately to emergencies, and when called to perform they may already be taxed by long hours or tedious procedures. Even when not actually on duty, doctors may be on call and prepared to perform services at a moment’s notice. Their duties therefore raise obvious and undeniable safety hazards. See AFGE L- *1473 2110 v. Derwinski, 777 F.Supp. 1493, 1499 (N.D.Cal.1991).

Defendants list some three hundred Bureau employees engaged duties related to medical care and administration. Of these, approximately one hundred employees appear to qualify as “physicians” or “dentists.” All of these employees are listed as primary law enforcement employees, i.e., employees engaged in duties at Bureau institutions. The privacy expectations of these employees are therefore diminished as noted above. Defendants have not, however, offered any evidence as to which of these employees are actually involved in the diagnosis or treatment of inmates or other Bureau employees at institutions on a regular basis. Nor do defendants appear interested in testing other medical employees whose duties may include regular diagnosis or treatment at institutions. The Court therefore holds that those licensed physicians and dentists in primary law enforcement positions who, in the regular course of their duties, must diagnose, treat, or directly supervise the diagnosis or treatment of patients on a daily or weekly basis may be subject to random urinalysis.

b. “Protection of Truly Sensitive Information.”

Defendants propose the random testing of some fifteen Bureau employees who have access to “sensitive” information. Ten of these employees are represented by plaintiff. Of these ten, four are secretaries or assistants with access to information about persons, some incarcerated in Bureau institutions, who cooperate with the federal government as witnesses or informants. Two others work in the Bureau’s Office of Internal Affairs, which investigates and monitors incidents of alleged staff misconduct. Another two work in the Bureau’s Correctional Services Branch, which is responsible for such issues as Bureau intelligence, investigation, and security and custodial arrangements. One evaluates witness program candidates. The last is a telecommunications specialist whose duties require extensive knowledge of the security systems within correctional facilities. Fourth Quinlan Decl. fllf 25-32. None of these employees are officially designated with “secret” or “top secret” clearance. .The two employees in the Office of Internal Affairs are “being processed” for top secret clearances; the two Correctional Services Branch employees were apparently “being reviewed” for submission for such clearance but are no longer under such consideration. Sixth Quinlan Decl. ¶ 16.

In Von Raab the Supreme Court acknowledged the government’s “compelling interest in protecting truly sensitive information.” 489 U.S. at 677, 109 S.Ct. at 1396. But the Court did not define the contours of “truly sensitive information.” See Harmon v. Thornburgh, 878 F.2d 484 at 491 . It merely upheld the testing of U.S. Customs officials who sought promotions to positions where they would be required to handle materials designated as “classified.” Id.; Von Raab, 489 U.S. at 677-78 , 109 S.Ct. at 1396-97 . Although Von Raab did not involve random testing, there is little doubt that under that decision the constitution permits the random urinalysis of government employees with access to “secret” or “top secret” information. Hartness v. Bush, 919 F.2d 170, 172-173 (D.C.Cir.1990), cert. denied, — U.S.-, 111 S.Ct. 2890 , 115 L.Ed.2d 1055 (1991) (permitting random urinalysis of executive agency employees with “secret” clearances); cf. National Treasury Employees Union v. Hallett, 756 F.Supp. 947, 951-52 (E.D.La.1991). The reason for this is the threat of “serious” or “exceptionally grave” damage to national security that a drug-impaired government employee with access to such information poses. Hartness v. Bush, 919 F.2d at 172-73 ; Harmon v. Thornburgh, 878 F.2d at 492 . Government employees with such access also have diminished expectations of privacy because of the sensitive nature of the information they handle and the processing they undergo for security clearance. See, e.g., Hartness v. Bush, 919 F.2d at 173 ; AFGE v. Cavazos, 721 F.Supp. at 1374-75 .

Defendants fear no potential threats to national security. Rather, defendants fear the potential breach of Bureau security, and the potential harm to individual wit *1474 nesses and informants that could result from disclosure of the information available to the designated employees. The safety of individuals who risk their lives to cooperate with the federal government is no trifling matter. Nor is the security of Bureau institutions and the need for strict confidence in matters relating to Bureau intelligence. Neither need, however, rises to the level heretofore approved as justifying random urinalysis. National security can be irreparably breached by a momentary lapse in alertness or judgment by one charged with its protection or secrets. Even the smallest such lapse by persons entrusted with secret or top secret information could be disastrous to our country’s well-being. In acknowledging the government’s compelling interest in protecting “truly sensitive information,” the Von Raab Court contemplated risks of this magnitude. See Von Raab, 489 U.S. at 677 , 109 S.Ct. at 1396-97 , citing Department of the Navy v. Egan, 484 U.S. 518 , 108 S.Ct. 818 , 98 L.Ed.2d 918 (1988) (upholding denial of security clearance to individual seeking job at repair facility for nuclear submarines). The Court also expressed concern that the category of persons to be tested under this rationale not be too broadly defined. 489 U.S. at 678, 109 S.Ct. at 1397. Defendants’ category, while perhaps well-intentioned, is too broad. The information sought to be protected is not so dangerous, and its disclosure not so immediately damaging, to justify the random urinalysis of the employees designated above. If and when any of these employees receive “secret” or “top secret” security clearances, defendants may petition this Court appropriately.

c. “Employee Integrity.”

Defendants attempted to justify all of their original testing program under this category. AFGE v. Meese, 688 F.Supp. at 554 . Their asserted interests under this rationale were preventing corruption within the Bureau and maintaining public confidence in it. This Court found neither of these interests compelling. Id. Defendants now propose to test only Bureau chaplains under the employee integrity rationale. They also assert that the integrity rationale adds further justification for the random testing of the other categories of Bureau employees whose proposed testing is discussed above.

Bureau chaplains counsel inmates and staff. Fourth Quinlan Deck ¶ 10. Counseling regarding the danger of illegal drug use is part of their duties. Id. At least one court has found that successful performance of drug counselors’ duties reasonably requires the counselors’ abstinence from illicit drug use. NFFE v. Cheney, 884 F.2d at 614. Drug counselors who use drugs, like drug interdiction agents who use drugs, may be “unsympathetic to their mission” because of their drug use. Id., citing Von Raab, 489 U.S. at 670, 109 S.Ct. at 1393 . As the D.C. Circuit has stated,

NFFE v. Cheney, 884 F.2d at 614. That court therefore allowed the random urinalysis of “direct service staff” in the Army’s substance abuse program. The court noted, however, that the employees on that staff were “primarily drug counselors.” Id. at 613 (emphasis added). Here, defendants state that drug counseling is “an important part” of the duties of Bureau chaplains but concede that it is only one aspect of their duties. Fourth Quinlan Decl. at ¶ 10. There is no showing that the “basic mission,” or even a major part of the duties of Bureau chaplains is counseling against drug abuse. Their very job titles suggest quite the contrary. Drug use by these employees would thus not appear to be “so dissonant with their responsibilities” to warrant their suspicionless testing. The testing of Bureau chaplains is therefore unwarranted.

Defendants assert that the interests of employee integrity provide further justification for the random testing of the Bu *1475 reau employees in the other categories addressed above. In this regard, defendants advance the same concerns presented in their initial motion, viz., maintaining public confidence in the Bureau and stopping the smuggling of illegal drugs into Bureau institutions. This Court was unmoved by these concerns the first time they were raised. AFGE v. Meese, 688 F.Supp. at 554 .

As for smuggling, the Court noted that while there was evidence that Bureau employees had been involved in bringing drugs into institutions, there was no proof that these employees were drug users. Id. Their testing would therefore not have served its intended purpose. The Court also noted that there were less intrusive measures for deterring drug smuggling. Id. As for “public confidence,” the Court found this the weakest of defendants’ concerns and no basis for the proposed testing. Id.

In Von Raab , the Supreme Court acknowledged the government’s compelling interest in ensuring that “front-line [drug] interdiction personnel” have “unimpeachable integrity,” and that drug users should be barred from positions directly involving the interdiction of illegal drugs. 489 U.S. at 670, 109 S.Ct. at 1393. Partially on this ground, one circuit court has upheld the annual urinalysis of a state prison’s corrections officers and supervisors who have “regular access to inmate populations” or “reasonable opportunity to smuggle drugs into the inmate population.” Taylor v. O’Grady, 888 F.2d 1189 , 1196-1197, 1199, 1201 (7th Cir.1989). Underlying that decision was the trial court’s finding of a “serious drug smuggling problem” within the prison at issue and a likelihood that correctional officers who smuggled drugs were likely to be chronic abusers. Id. at 1193, 1197; see also McDonnell v. Hunter, 809 F.2d 1302 , 1308 (8th Cir.1987).

There is no denying the serious threat to prison security, and thus the compromise of the basic mission of the prison, caused by employees’ smuggling of drugs to prisoners. See, e.g., Taylor v. O’Grady, 888 F.2d at 1197. Defendants have once again presented evidence of numerous incidents of drug smuggling to inmates by employees at Bureau institutions. See Third Quin-lan Dec. at 116. Again, however, there is no significant nexus between these employees’ drug smuggling and their drug use 5 Testing employees for drug use would thus serve only marginally, at best, the government’s interest in preventing drug smuggling. Nothing in the recent precedents supports testing in such circumstances. 6 Nor, it might be noted, have defendants specified which employees would be encompassed by the drug smuggling rationale. This Court specifically asked defendants to identify which of the listed employees had “regular (daily or weekly) contact” with inmates. In their response defendants declined such specificity, stating only that primary law enforcement employees are “regularly subject to contact with inmates” and secondary law enforcement employees have “only periodic contact” with them. Sixth Quinlan Deck at 111113-14. There is thus no clear justification for testing any • *1476 Bureau employees under the drug smuggling rationale.

Nor. is the government’s generalized interest in the integrity of its work force a sufficient justification for testing all the proposed Bureau employees. See Taylor, 888 F.2d at 1196. The integrity interest does, however, justify the testing of certain employees. All Bureau employees share dual responsibilities of “maintaining safe and secure institutions” and “modeling society’s mainstream values and norms.” Exh. 1 to Fourth Quinlan Deck Neither of these goals is met by drug-impaired employees whose duties require frequent or prolonged contact with inmates. As noted above, defendants decline to specify which primary law enforcement employees have such inmate contact. Rather, defendants repeatedly emphasize the “correctional workers first” concept in which all Bureau employees are trained. See id.; Exh. D to Sixth Quinlan Decl. Again, this rationale would embrace all listed Bureau employees and is' far too broad. The Court holds that the interests of employee integrity justify the testing of only those employees in primary law enforcement positions who (1) have direct contact with inmates (2) on a daily or weekly basis (3) for periods of one hour or more each day of contact. “Direct contact” means physical proximity; this could come through the supervision of inmates (as with corrections officers on cell patrol) or through non-supervisory consultation (as with doctors, counselors, or teachers). Sporadic and unprolonged inmate contact— such as by telephone operators, plumbers, painters, and myriad other positions identified by defendants — would not justify random testing under this rationale.

It deserves noting that this testing designation, while somewhat unique, is narrower than the “direct contact” standard apparently conceded by plaintiffs and upheld in Taylor, 888 F.2d at 1197; see also McDonnell, 809 F.2d at 1308. It might include Bureau chaplains, while also meeting the bulk of defendants’ concern' for testing those employees with opportunities to smuggle drugs to inmates. As with the other categories designated above, it is defendants’ responsibility to interpret and apply this standard, and to test under it only in accordance with the letter and spirit of the Court’s ruling.

C. Accident and Unsafe Practice Testing

The Bureau program would test employees who “apparently cause” accidents on the job or who, while on duty, engage in unsafe job-related activities that pose a danger to others or to “the overall operation of the Bureau.” Program Statement at 11, Art. VII. Whether or not to test an employee under this rationale would be decided by employee supervisors, but a decision to test an employee would require that the accident or unsafe practice (1) involves personal injury or death or (2) results in property damage of more than $2,000, or that (3) the same employee has more than one accident or unsafe practice during a twelve-month period. Id. Plaintiffs protest that the personal injury and monetary damage thresholds to trigger such testing in, respectively, the first and second of these categories are too low. They also contend that the third category, by setting no minimum threshold of injury and failing to define “unsafe practice,” improperly grants unfettered discretion to local officials.

There is little doubt that the Bureau’s interest in determining the cause of accidents and otherwise ensuring the safety of its employees constitutes a “special need” justifying post-accident testing absent a warrant or probable cause. See Connelly v. Newman, 753 F.Supp. at 295-96 . This Court must therefore balance this interest against the reasonable expectation of privacy of Bureau employees. That privacy interest has already been discussed at length supra. The Court will therefore focus its analysis on the Bureau’s interest and the specific testing categories.

Recent cases provide considerable guidance. In Skinner, the Supreme Court approved the mandatory post-accident testing of railroad employees directly involved in accidents or incidents of certain magnitudes. 489 U.S. at 609, 109 S.Ct. at 1408- *1477 09. The injury thresholds for such testing were $50,000 or a “reportable injury” in “impact accidents,” and $500,000 or a fatality in “major train accidents.” 7 M 8 In Teamsters, the Ninth Circuit upheld the mandatory testing of truck drivers after accidents involving a fatality, an injury “demanding immediate medical treatment away from the scene of the accident,” or at least $4,400 in property damage. 932 F.2d at 1308 . Relying on Teamsters, this Court rejected the identical thresholds proposed here as too low and subject to local discretion. See AFGE L-2110 v. Derwinski, 777 F.Supp. at 1501-02; see also AFGE Local 1533 v. Cheney, 754 F.Supp. at 1409 (enjoining post-accident testing of Navy employees in program where local officials had discretion to set triggering criteria as to personal injury or monetary damage thresholds); Connelly v. Newman, 153 F.Supp. at 295-97 (enjoining Office of Personnel Management post-accident plan with no causation requirement and thresholds of “death or personal injury requiring immediate hospitalization” or damage in excess of $1,000); National Treasury Employees Union v. Yeutter, 733 F.Supp. 403, 406, 416-17 (D.D.C.), affd on other grounds and vacated in part, 918 F.2d 968 (1990) (upholding testing of USDA employees “apparently causing” incidents involving fatality, hospitalization, or property damage over $10,000).

Skinner and Teamsters involved industries that were both highly regulated and fraught with the daily discharge of duties whose dereliction, even for a second, could cause tremendous property damage or take lives. Connelly , on the other hand, involved a civil service office without these two crucial characteristics. The Bureau of Prisons, with employees who may carry firearms and are responsible for maintaining safe and secure prisons, lies somewhere between these two poles. The Bureau does have an interest in determining (and deterring) the possible role of drug use in harmful accidents and practices. Again, however, this interest is compelling only at Bureau institutions. Some post-accident testing plan for primary law enforcement employees therefore appears warranted. Considering both Teamsters and the wide variety of activities that Bureau employees might be engaged in, the $2,000 threshold for testing under the second category is appropriate. The absence of a threshold of personal injury in the first category, however, leaves too much discretion to local officials. It also allows testing for the most minor of injuries, including, as plaintiffs point out, the dropping of a stapler on a foot. In accordance with Teamsters, testing in that category must require a minimum level of personal injury. The third category of post-accident testing is incurably vague. Defendants concede that it would allow testing at lower thresholds of injury than the first two categories. Testing under that category must therefore be enjoined.

Accordingly, the Court holds that the Bureau may test those employees in primary law enforcement positions who apparently cause accidents or engage in unsafe practices (1) involving personal injury that requires immediate medical treatment, or (2) resulting in more than $2,000 damage.

D. Reasonable Suspicion Testing 9

The preliminary injunction issued by this Court enjoined all Bureau testing *1478 except of an employee as to whom there was reasonable suspicion that “drug use by that employee impairs his or her ability to perform official duty.” 688 F.Supp. at 556 . Under its modified program the Bureau proposes to test under this category if reasonable suspicion warrants “rational inferences that a person is using drugs.” Program Statement at 11. For employees in positions designated for random testing, the suspicion may apply to “conduct either on or off duty.” Id. For all other employees, the suspicion must be of “on-duty illegal use or impairment.” Id. Plaintiffs challenge this testing since it allows testing for off-duty impairment. They also propose specific criteria for a finding of reasonable suspicion.

Once again the Court balances the interests at stake. In Von Raab and Skinner, the Supreme Court recognized a government interest in off-duty drug use only with respect to those employees with extraordinary duties — e.g., drug interdiction, firearm use, access to truly sensitive information, and railroad work. See Von Raab, 489 U.S. at 670-71, 676-77 , 109 S.Ct. at 1393, 1396-97 ; Skinner, 489 U.S. at 632-33 , 109 S.Ct. at 1421-22 ; NTEU v. Yeutter, 918 F.2d at 974-75 . This Court has found such extraordinary duties in only certain categories of Bureau employees proposed for random testing. The compelling interest in detecting and deterring off-duty drug use or impairment outweighs the privacy interests of only these employees. See NFFE v. Cheney, 742 F.Supp. 1, 3 (D.D.C.1990). Thus, the Court holds that reasonable suspicion of off-duty drug use or impairment can serve as a ground for testing only (1) employees in primary law enforcement positions who, in the course of their regular duties, are issued or given access to firearms for use on a daily or weekly basis; (2) licensed physicians and dentists in primary law enforcement positions who, in the regular course of their duties, must diagnose, treat, or directly supervise the diagnosis or treatment of patients on a daily or weekly basis; and (3) employees in primary law enforcement positions who (a) have direct contact with inmates (b) on a daily or weekly basis (c) for periods of one hour or more each day of contact. All other Bureau employees may be tested only where there is reasonable suspicion of on-duty drug use or impairment.

The Court’s present ruling does not conflict with the recent holdings of this Court in Connelly, 753 F.Supp. at 297-99 , and AFGE Local 1533 v. Cheney, 754 F.Supp. at 1426-27 . In both of those cases, testing based on off-duty behavior was deemed justifiable because “reasonable suspicion” was defined by five specific criteria that made such testing relevant to the determination of on-duty impairment. These criteria do not appear in the Bureau’s present program. Moreover, today’s ruling accords with the more recent, decision of this Court in AFGE L-2110 v. Derwinski, 777 F.Supp. at 1501 (allowing reasonable suspicion testing for off-duty conduct of only those employees in “safety-sensitive” positions).

Plaintiffs ask the Court to provide two criteria to define what would constitute reasonable suspicion to justify testing. The proposed criteria are: (1) evidence of specific, personal observations concerning job performance, appearance, behavior, speech, or bodily odors of the employee; or, if the suspicion is based on hearsay evidence, (2) corroborative evidence from a manager or supervisor with training and experience in the evaluation of drug-induced impairment. The Court finds both criteria warranted and appropriate. Despite defendants’ assertion to the contrary, the Bureau’s history of testing under this category has not proven as reasonable as in Connelly . There, with the help of five specific criteria, the Office of Personnel Management tested one employee over some eighteen months. 753 F.Supp. at 298 . Here, the Bureau admits to having tested thirty-three employees in some thirty-six months. Fifth Quinlan Decl. at ¶ 2. Only eleven of these tests reported positive for illegal drug use. Id. While the number of employees tested under this category in *1479 that period may have been uncharacteristically large since this Court had enjoined testing under all other categories, some guidance to the Bureau appears appropriate. The Court therefore holds that “reasonable suspicion" to test within the parameters outlined above must be supported by (1) evidence of specific, personal observations concerning job performance, appearance, behavior, speech, or bodily odors of the employee; or, if based on hearsay evidence, (2) corroborative evidence from a manager or supervisor with training and experience in the evaluation of drug-induced impairment. 10

Accordingly,

IT IS HEREBY ORDERED that defendants’ motion for summary judgment is GRANTED in part and DENIED in part, and defendants may implement their drug-testing program subject to the limitations of the following permanent injunction, which is hereby entered as the final judgment in this case:

PERMANENT INJUNCTION

1. Random Testing. The Bureau of Prisons may conduct random urinalysis testing only of:

2. Post-Accident Testing. The Bureau may conduct urinalysis testing of those employees in primary law enforcement positions who apparently cause accidents or engage in unsafe practices (1) involving personal injury that requires immediate medical treatment, or (2) resulting in more than $2,000 damage.

3. Reasonable Suspicion Testing. The Bureau may test those employees of whom the Court has approved random drug testing, upon reasonable suspicion of on-duty or off-duty drug use or impairment by these employees. All other Bureau employees may be tested only where there is reasonable suspicion of on-duty drug use or impairment. “Reasonable suspicion” to test in either circumstance must be supported by (1) evidence of specific, personal observations concerning job performance, appearance, behavior, speech, or bodily odors of the employee; or, if based on hearsay evidence, (2) corroborative evidence from a manager or supervisor with training and experience in the evaluation of drug-induced impairment.

4. No other aspect of the Bureau’s drug-testing plan is affected by this injunction.

5. This injunction is binding upon defendants and their officers, agents, servants, employees, and attorneys, as well as those persons in active concert or participation with them who receive actual notice hereof by personal service or otherwise.

6. No person who has notice of this injunction shall fail to comply with its letter and spirit, nor shall any person subvert its letter or spirit by any sham, indirection, or other artifice.

7. The Court retains jurisdiction to modify this injunction at any time and from time to time, upon its own ‘motion or the motion of any party, in the interest of effectuating its intendments or furthering the ends of justice under all applicable law.

. Plaintiffs do not challenge the testing of job applicants, probationary employees, and management employees, nor do they challenge follow-up or voluntary testing. Plaintiffs have also withdrawn their challenge of the program under the Civil Service Reform Act.

. In making this determination the Court relied on Railway Labor Executives’ Ass’n v. Burnley, 839 F.2d 575 (9th Cir.1988) ("RLEA /’), which was reversed by Skinner.

. The Bureau has, moreover, had job applicant testing since 1984. First Quinlan Decl. at ¶ 5(b).

. Defendants have provided clear evidence of only one incident in the past five years in which an emergency situation necessitated the issuance of firearms to Bureau employees who would not otherwise carry them. Fourth Quin-lan Decl. at ¶ 24; Sixth Quinlan Deck at ¶ 11 and Exh. E. Only two such employees are specifically referred to with regard to this incident. Fourth Quinlan Deck at ¶ 24. For at least one of these employees, it was apparently the only time he was so issued a firearm in sixteen years of employment at Bureau offices and institutions. Exh. E to Sixth Quinlan Deck This employee had, moreover, received firearm training and had worked as a correctional officer at one Bureau institution. Id. at ¶¶ 1, 5.

. Defendants’ evidence includes a few purported examples of such a connection, but most of these examples are far from conclusive. See, e.g., Third Quinlan Dec. at ¶ 6(a), (b), (e), (i), (w).

. In Taylor, the district court had expressly found a strong correlation between prison employee drug smuggling and use. 888 F.2d at 1197; Taylor v. O'Grady, 669 F.Supp. 1422, 1429, 1437 (N.D.Ill.1987).. In McDonnell, the court simply declared it "logical to assume that those who use the drugs, and who come into regular contact with the prisoners, are more likely to supply drugs to the inmates.” 809 F.2d at 1308.

It is true that the recent Supreme Court and Ninth Circuit decisions have held that the absence of a demonstrated drug use problem does not defeat the substantial government interest in preventing such a problem from developing in certain contexts. Von Raab, 489 U.S. at 674 -75 & n. 3, 109 S.Ct. at 1395-96 & n. 3 (Customs service); Teamsters, 932 F.2d at 1305 (trucking industry); IBEW, 913 F.2d at 1461 (pipeline industry); cf. Bluestein, 908 F.2d at 456 (aviation). In these contexts, however, as in the present case, testing is allowed despite the absence of a demonstrated drug use problem since it quite obviously and directly deters drug use. There is no such obvious or direct connection between testing for drug use and deterring drug smuggling.

. The Court did not specify what constitutes a "reportable injury."

. Testing was also mandatory after any accident involving a fatality to any on-duty employee. 489 U.S. at 609, 109 S.Ct. at 1408-09. Testing was authorized, but not mandatory, where a supervisor had reasonable suspicion of an employee’s contribution to an accident or incident, and if an employee violated certain rules such as those regarding compliance with signals and speeding. Id. at 611, 109 S.Ct. at 1409.

.Defendants protest that this aspect of the Bureau program is not properly before the Court since the complaint in this case contains no paragraph or phrase objecting to reasonable suspicion testing. To the extent that this point has merit, it should be noted that plaintiffs have objected to this aspect of the Bureau program in their present opposition papers and subsequent correspondence with the Court, and defendants have had ample opportunity to respond. Moreover, this aspect of the program must be ad *1478 dressed to conform it with the remainder of the Court's judgment.

. The Bureau has already stated its commitment to providing such managers and supervisors. Program Statement at 5.