The decision is believed to be the first to overturn a conviction based on a challenge to the FBI analysis of the lead content of bullets since the National Academy of Sciences last year raised new questions about the technique the FBI has used for decades to match bullets to crimes.

FBI Lab Director Dwight Adams asked for the academy study in 2003 after a retired bureau metallurgist began questioning the validity of the science that matches bullets by comparing the chemical composition of their lead content. The academy, chartered by Congress and privately run, is widely respected.

The FBI request came as the Associated Press reported the retired metallurgist's findings based on his own research into the science behind the lead bullet analysis. The AP also reported that an FBI scientist had admitted giving false testimony about lead bullet analysis in a Kentucky case.

Adams has estimated that the technique has been used in about 2,500 cases since 1980, and has been mentioned in court testimony about 500 times since then.

William Tobin, the retired FBI metallurgist who first questioned the technique, had submitted a sworn statement in the New Jersey case that resulted Monday in a new trial for Michael S. Behn, who was sentenced to life in prison after his 1997 conviction in the shooting death of a coin dealer.

The FBI is the only law enforcement agency that analyzes the metal content of bullets. It is done when bullet fragments are too small or damaged to compare the marks left on the slug by the barrel of the firearm. The goal is to determine if the bullet from the crime matches other bullets found in the suspect's possession or weapon.

In a technique known as chaining, researchers compare the amounts of trace elements in a series of bullets in a box. If they find that bullet A is like bullet B and B is like C and C is like D and so on, they then conclude that A is the same as E because they are part of the same chain.

In the New Jersey case, the appellate court said the FBI analysis that used chaining to link bullets found at Behn's residence with those used in the killing was the only expert testimony in the circumstantial case that was not countered by the Behn's lawyers during his trial.

"We conclude that the expert testimony was based on erroneous scientific foundations," the court said.

Behn's lawyer had asked the appeals court to order a new hearing on the evidence challenging the FBI analysis. But the court went a step further, reversing his conviction and ordering a new trial.

"The integrity of the criminal justice system is ill-served by allowing a conviction based on evidence of this quality, whether described as false, unproven or unreliable, to stand," the judges said.

Barry Scheck, president of the National Association of Defense Lawyers, said the ruling will probably be followed by similar challenges in other courts.

"There's no question since the National Academy of Science report that testimony pursuant to CBLA (chemical bullet lead analysis) is unreliable," he said. "You can't use it for any probative evidentiary purpose. In many of these cases, people been wrongly convicted."