Recent documents obtained establish a 1994 lawsuit was filed against Lilly for victims of a Louisville workplace shooting. The gunman, who killed himself and eight others, had been treated with Prozac.

Teicher, a prominent researcher and clinician, was the first to publish case reports showing an apparent link between Prozac and suicidal behavior in adults. But a few months before his March 1990 report, Teicher said he asked Lilly officials if studies showed such a link.

" 'Oh no, no, we never heard of such a thing,' they told me," Teicher said. But studies from the 1980s showed such dangers, he said, and German drug regulators wouldn't license the drug when Lilly first applied in 1985, citing "suicidal risk." The Food and Drug Administration (FDA) approved Prozac in 1987. But when problems started surfacing, the agency held a scientific advisory panel hearing in 1991 and found no evidence of a link to suicide.

Before that hearing, Lilly published studies showing Prozac was safe, Teicher said. "They culled patients from their worldwide trials, they cherry-picked the studies, leaving out the trials showing problems."

Lilly spokesman Morry Smulevitz said "we are not aware of the conversation that Dr. Teicher refers to with Lilly officials." He also denied that the company suppressed negative evidence about the drug. "Lilly believes in full and appropriate disclosure of clinical trial data," Smulevitz said.

The documents summarized in the BMJ article suggest that twice as many patients on Prozac as on a placebo may experience such symptoms as anxiety, agitation and nervousness — 38% versus 19%. These symptoms are important because therapists say they can precede suicide or violent acts.

Another document appears to show an FDA official questioning Lilly's actions in deciding apparent suicide attempts in studies weren't real attempts or were not due to the antidepressant. Scientists have some discretion in identifying the causes of bad reactions in a study.

David Graham, associate director in the FDA's Office of Drug Safety, reviewed Prozac's safety record in 1990. "Lilly excluded so many cases of suicidal behavior that I felt a problem couldn't be ruled out," he said.

The FDA is still reviewing the BMJ documents, FDA spokeswoman Susan Cruzan said. "But to date we have not seen anything that would lead us to question the conclusions of the (1991 scientific advisory) committee" that there was no tie between the drug and suicide attempts.

Teicher, who considers Prozac valuable, said many of the problems with suicidal behavior were in patients given high doses, and that's how the drug was used for the first few years in the USA. "American people were guinea pigs for a few years. If we had known the truth, we would have used it more wisely from the start," Teicher said.