Two months before his arrest, Wright was discovered to have stored an assault rifle, a pistol, and 1,000 rounds of ammunition in his barracks. Although he was demoted a rank, his PRP certification was not revoked. He was allowed to keep the weapons, which he subsequently used in the murder. The Army was so anxious to cover up the case, Wright says, that it granted him an honorable discharge and a Good Conduct Medal while he was in jail awaiting trial on charges of first-degree murder. His PRP certification was not revoked until January 1988, 11 months after his arrest.

After Metcalf's arrest, the Navy reviewed his PRP screening and declared that there was nothing that could have tipped off investigators to his potential for violence. However, a copy of the screening obtained by Mother Jones suggests a number of conspicuous warning signs.

During his PRP screening in 1986, Metcalf claimed that his mother was dead. When military investigators learned that she was still alive, Metcalf changed his story, saying that he had broken off relations with her due to a financial dispute. Although investigators traveled to Metcalf's hometown of Redfield, Arkansas, to interview a former neighbor, they did not interview his mother.

A former acquaintance told investigators that Metcalf had been questioned by police in the unsolved 1981 murder of his then-girlfriend, Rita Sanders. Metcalf denies any role in Sanders' death and says, "It was never brought up in any of my interviews. They never said anything about it." The Navy didn't even bother to contact the police in Oxnard, California, who considered Metcalf the prime suspect in the murder. (The case was eventually dropped for lack of evidence.) Meanwhile, according to his PRP file, Metcalf married a woman named Barbara Adsett four days after Sanders' murder. The couple divorced four months later and Adsett, claiming Metcalf had beaten her, obtained a restraining order to keep him away from her and her family.

During Metcalf's PRP screening, a number of his past acquaintances told the Navy that he was not trustworthy. William Hill, his former supervisor at a machine shop called C.E. Miller, told investigators that Metcalf stole $200,000 worth of tools and welding equipment from the firm. "Hill further advised that [Metcalf] likes material things and will do anything for money, and would be easily tempted to sell sensitive material," reads Metcalf's file. Another former co-worker from C.E. Miller told PRP screeners that Metcalf used amphetamines and had once appeared at his house with "a bundle of $10,000 in $100 bills." Yet another former acquaintance called Metcalf a "habitual liar," who "fantasizes continuously."

None of this stopped the Navy from approving Metcalf for PRP duty. Nor did later monitoring detect anything unreliable about Metcalf. His last performance evaluation, completed just months before the murder of the Sawyers, called him a "key member of the Strategic Fire Control Division," whose "outstanding technical abilities and 'can do' attitude have earned him the respect of all who work with him." An evaluation from a few years earlier was even more exuberant, calling Metcalf a model to be "emulated by all."

In a chilling videotaped interview conducted by the National Security News Service, Metcalf claimed that he was frequently left alone in the Fire Control Center aboard the Alaska, and talked about how easy it would have been to bypass safeguards and single-handedly launch a nuclear strike with the submarine's intercontinental ballistic missiles. "In the time it takes to get everyone in the response team together, someone who knows what they're doing could already be launching missiles.... One person is all it takes."

Skip Beard, a former commanding officer on a nuclear submarine and a technical consultant to the movie Crimson Tide, says Metcalf's scenario is nonsense. "The possibility of overriding [safeguards] is so remote as to not exist," he says. "He's blowing hot air." Still, the very fact that Metcalf spent an enormous amount of time theorizing a personal first strike is itself deeply disturbing. And Beard concedes that a PRP-certified worker aboard a ship could ignite the fuel or explosives in a missile by firing into it with a gun. "If that happened," he says, "everybody on the boat is dead."

A review of dozens of files of people who were decertified from PRP during the early 1990s, obtained under the Freedom of Information Act, also calls into question the "stable and dependable" nature of PRP personnel. One person was kicked out after being "overcome by a severe emotional disturbance which caused him to lose his ability to communicate.... He was subsequently diagnosed with having a personality disorder which is a deeply ingrained, maladaptive pattern of behavior." Another was decertified after "an alcohol-related incident at a local bar where he allegedly assaulted a civilian, who was hospitalized with severe head injuries." A third case involved a soldier who tried to suffocate his 12-day-old daughter when she wouldn't stop crying. Other PRP-approved personnel tried to commit suicide or were found guilty of crimes ranging from rape to burglary; one man was found drunk while on duty with a bottle of Jack Daniels concealed in his waistband.

Mother Jones learned of further problems with the program from two former PRP-certified men, both of whom are currently serving lengthy prison terms for offenses committed while they were still in the military. Both asked that their names not be used, but provided documentation to support their accounts. One, who served as a sonar technician on a nuclear missile submarine, was expelled from PRP when he tested positive for cocaine during a random drug test. He subsequently deserted and was later arrested for a spree of crimes ranging from bank robbery to methamphetamine use. He claims that abuses that should result in one's expulsion from PRP are frequently ignored by certifying officials. "Military men are prone to hard drinking, but this is routinely overlooked," he says. "I can tell you I never turned myself in, or any of my peers, for drinking binges."

The second person -- a Marine expelled from PRP in 1993 and now serving a 15-year prison term for indecent assault and sodomy -- guarded nuclear weapons at a naval base. He, too, claims that military officers overlooked problems such as heavy drinking or depression. In one case, he says, staff shortages were temporarily solved by having both a man who had been removed from PRP duty due to a medical problem and an uncertified administrative clerk called in to guard nuclear weapons. "Manpower [demand] at special weapons stations far exceeds the number of qualified Marine personnel," he says. "As a result, waivers are issued, faults are ignored, and unqualified people sit [at PRP] posts."

Col. Landis concedes that the system is not foolproof, noting that the overall decertification rate for PRP-approved personnel is 3 to 4 percent. "Like society, we're going to have people of all types in the services," he says. "We live in a violent environment." Landis admits that investigators sometimes fail to uncover all relevant information on a PRP candidate, but says that if damaging information is found, the person will not be certified. (Landis declined to comment on why Metcalf was granted PRP status, saying that he, Landis, was not working with the program at the time.) In general, says Landis, PRP is a "great program."

Critics are less sanguine. "They're not looking for problems, and every time you start looking you find problems," says a source with firsthand knowledge of PRP. Postcertification monitoring is also impeded by service members' reluctance to inform on erratic behavior by their peers. A 1992 report from the General Accounting Office to then-Defense Secretary Richard Cheney noted that such reporting "was rarely a factor in alerting supervisors to potential problems."

Stanford's Abrams recommends that the Pentagon strengthen PRP by requiring a physician to examine all candidates, using standardized psychological testing, and improving its postapproval monitoring procedures. Others say the entire program should be subject to tough independent scrutiny. The Department of Energy reviewed personnel problems at its nuclear weapons and research complexes, and published the findings in an April 2, 1993, report. In one case, an employee stole uranium oxide from one site in an apparent attempt to extort money from the federal government. In another, the DOE disclosed "several tampering incidents" at the Los Alamos National Laboratory's main plutonium facility, in which a worker or workers had loosened valves in a hydrogen feed line. If the tampering had gone undiscovered, says the report, possible consequences would have included "a fire-induced or explosion-induced release of plutonium and other radioactive or toxic material to both on-site and off-site environment."

Scott Sagan says there's no reason to believe that similar problems don't exist at the Pentagon's nuclear weapons operations. "Until the Defense Department makes public its recent records on PRP, the U.S. public won't know how grave the current risk is," he says.

The risk was illustrated dramatically in mid-September in the port city of Murmansk, Russia, when a Russian sailor wielding a machine gun single-handedly commandeered a nuclear submarine, and then fatally shot eight of his fellow crew members before turning the gun on himself. And while it might be only a coincidence, it's worth noting that the Pentagon has been helping the Russian military enhance its own personnel reliability system.

The National Security News Service, which contributed to this report, is a nonprofit organization that investigates the Pentagon and the arms industry.