The FBI said Thursday that it had discontinued the use of bullet-lead matching, a forensic technique used for at least 25 years that had been heavily criticized as inaccurate and misleading.
The bureau suspended its use in 2004 after a report by the National Research Council found the technique could be "seriously misleading" and "objectionable."
The council's finding called into question FBI testimony in hundreds of cases involving murder and other serious crimes.
"It's a victory for good sense and good science over the kind of nonsense the FBI was representing in court," said William C. Thompson, a professor of law and criminology at UC Irvine.
The FBI said it would alert about 300 courts and prosecutors that since 1996 had received bullet-lead laboratory reports indicating positive results.
"These agencies may take whatever steps they deem appropriate, if any, given the facts of their particular case," said an FBI statement released Thursday.
Overall, the bureau estimated it had conducted 2,500 bullet-lead examinations for federal, state, local and foreign cases since the early 1980s.
In bullet-lead analysis, crime-scene bullets are tested for trace elements, such as antimony, silver and tin. Examiners then compared those elements to levels found in bullets in a suspect's possession.
FBI examiners had testified that crime-scene bullets could be linked to bullets found in a box owned by a suspect. The method was often used when no gun was found, making it impossible to identify a bullet by matching rifling marks left by a gun's barrel.
FBI Laboratory Director Dwight E. Adams said the agency, after conducting its own evaluation over the last 14 months, concluded that the manufacturing and distribution of bullets was too variable to make the matching reliable.
"It wasn't possible to obtain accurate, easily understandable probability estimates like it is with other methods, such as DNA," Adams said.
Nonetheless, he said, "we stand by the results of the reports we have already issued."
FBI testimony had sometimes gone beyond the lab reports, and juries had often accepted such testimony as important in criminal cases.
A Times investigation published in 2003 suggested the FBI might have exaggerated the scientific validity of bullet-lead matching.
In one case, that of Michael Behn, a New Jersey man convicted of murder in 1997, an FBI examiner testified that bullets collected from the victim's body matched those in Behn's home. In Behn's case, Adams blamed the prosecutor, who "overstated the evidence," he said.
Behn challenged the bullet-lead evidence on appeal. His conviction was overturned in March, and he was granted a new trial. His case is pending.
Thompson said the FBI should be more aggressive in making sure wrongful convictions due to faulty bullet-lead evidence were overturned.
"If the FBI is serious about correcting the problems, they should make the list of cases known to organizations such as the National Assn. of Criminal Defense Lawyers that might be able to provide assistance to incarcerated individuals," he said.