The US Department of Energy (DOE) has said that it plans to continue routine polygraph testing of scientists who have access to classified information in its nuclear-weapons laboratories.

The decision has angered many of the researchers, who say that it ignores a National Academy of Sciences study published last October, which found that polygraph testing has “extremely serious limitations” in discriminating between honest employees and likely spies.

Announcing the plan, energy secretary Spencer Abraham cited both the war in Iraq and the war on terrorism as reasons for continuing with the polygraphs. “As the steward of the nation's nuclear-weapons stockpile, the department has an obligation to use the best tools available to protect its most sensitive information,” he said in a statement on 14 April.

But critics say that the National Academy study found polygraphs to be an unscientific tool that is unlikely to detect spies. “It's lore, it's theology, and unfortunately, the secretary of energy has bought it,” says Alan Zelicoff, a bioweapons specialist at the Center for National Security and Arms Control at Sandia National Laboratory in Albuquerque, New Mexico. “I can't tell you the number of e-mails I've got today saying, 'This represents a new low',” he adds.

The DOE introduced routine polygraphs for thousands of staff at its nuclear-weapons labs three years ago as a response to the case of Wen Ho Lee, an engineer at the Los Alamos National Laboratory in New Mexico, who was charged with passing nuclear secrets to a foreign power. Lee was not convicted, but pleaded guilty to lesser charges.

Counterintelligence officers at the department argued that polygraphs would help them unearth potential spies, but many scientists pointed to a lack of data that the tests actually work.

Their scepticism led Congress to commission the National Academy study, which found that polygraphs could sometimes detect deceitful responses to directed questions, but lacked the precision to identify lies in response to the sort of open-ended questions asked in security-clearance interviews. The study concluded that testing left officials with an “unacceptable choice” between falsely judging too many loyal employees or leaving security threats undetected.

Despite these conclusions, the DOE plans to continue the tests. “It would not be appropriate to downgrade one of the most effective tools we have,” says DOE spokeswoman Jeanne Lopatto. She adds that even if the study questioned the polygraph's utility, it still suggested that the tool can be effective if used as a trigger for a detailed follow-up investigation.

The study's authors disagree. “The report is clearly taken out of context and distorted,” says Stephen Fienberg, a professor of computer science and statistics at Carnegie Mellon University in Pittsburgh, Pennsylvania, who chaired the National Academy study. “There is virtually no science behind the screening process on which they rely,” he adds.

Politicians representing the areas around the weapons labs are also unconvinced. In a statement, Congresswoman Ellen Tauscher (Democrat, California) said that she was “shocked” by the decision and called for hearings. Senator Pete Domenici (Republican, New Mexico) and Jeff Bingaman (Democrat, New Mexico) also voiced concern at the plan, and may hold a hearing after a period of public comment on it ends on 13 June.

“I'm disappointed by the decision,” says Jeff Colvin, a physicist at Lawrence Livermore National Laboratory in California and an officer for the lab's union for scientists and engineers, which opposes the screening. “It's clear that the DOE will not be swayed by logical arguments or scientific facts.”