The best treatment for depressed adolescents is a combination of Prozac and talk therapy, although the antidepressant does carry a risk that some patients might harm themselves or others, concluded a long-awaited study released yesterday.

Although the findings may seem contradictory, the study could begin to address the roiling controversy over the safety and effectiveness of antidepressants in children: Drugs such as Prozac might increase the risk that a minority of adolescents will harm themselves or others, even as they reduce depression among the majority.

The results will be weighed by regulators at the Food and Drug Administration, who have scheduled a meeting in September to discuss analyses of antidepressant trials. Two previous agency analyses found increased signs of suicidal behavior among children taking the drugs. British authorities warned doctors last year not to prescribe most drugs called selective serotonin reuptake inhibitors to depressed children.

The new study was eagerly anticipated by doctors, parents and regulators: Virtually all clinical trials of antidepressants to date have been conducted by pharmaceutical companies. Although several researchers who conducted the new study have done work for the pharmaceutical industry, the new study was paid for entirely by taxpayers.

Thomas Insel, director of the National Institute of Mental Health, which paid $17 million for the study, said the study strongly tilted the risk-benefit ratio in favor of employing Prozac to treat depression. At the same time, Insel said, the increased risk of harm among children who got Prozac was worrisome, and he indicated that the drug had to be used cautiously.

"You can't write a prescription for Prozac the way you write a prescription for penicillin," he said in an interview last week. "There is no question there is an increase of adverse events on the drug -- there is more than a twofold increase in adverse events in those who are on the drug versus those who are not. That's a lot."

Critics of the drugs have seized on the signs of risk to demand the medications be more tightly regulated or even banned. Congressional investigations are underway into whether drug companies deliberately concealed information about dangerous side effects.

Most American psychiatrists, who generally favor the drugs, point to the risks of leaving depression untreated. Yesterday's study indicated that Prozac was effective in treating depression and that it was superior to a form of psychotherapy called cognitive behavior therapy. Combining the medication with cognitive behavior therapy produced the best results.

The study hinted that adolescents who got both drug and talk therapy had a lower risk of harming themselves or others compared with those who got the drug alone.

The challenge for clinicians and researchers, Insel said, was to try to tease apart the subgroup of children who might be adversely affected from those who could be helped.

Although this particular study recruited a broader range of patients than trials conducted by drug companies, Insel and other researchers cautioned that adolescents at high risk of suicide were excluded. Questions have grown about whether such patients are the most likely to be helped by antidepressants or the most likely to harm themselves.

In the study published in this week's Journal of the American Medical Association, seven adolescents among the 439 in the study attempted suicide during the 12-week trial. Four were getting Prozac and cognitive behavior therapy, two were getting Prozac alone, and one was receiving cognitive behavior therapy alone.

Insel, Susan Silva, an associate research professor of psychiatry and behavioral sciences at Duke University, and Richard Glass, a psychiatrist at the University of Chicago and deputy editor at JAMA who wrote an editorial accompanying the study, all warned against drawing broad conclusions from the small numbers of adolescents who harmed themselves or others.

Two internal FDA analyses of the pharmaceutical industry trials have suggested an increased risk of suicidal behavior among children getting antidepressants. But taken individually, the industry studies have been inconsistent, and pooling their results has proven complicated. Researchers have also struggled with whether harmful behaviors such as self-mutilation should be classified as suicide attempts. No child has committed suicide in any of the trials.

"At this point, the data is still equivocal and raises many more questions than it answers," said David Fassler, a child psychiatrist who spoke on behalf of the American Psychiatric Association. "We need to take any risk of suicide very seriously, but we also need to weigh it against the risk of suicide in kids who are not treated."

The study showed that about 61 percent of depressed 12- to 17-year-olds getting Prozac -- and 71 percent of those getting Prozac and cognitive behavior therapy -- improved significantly. About 43 percent of adolescents getting cognitive behavior therapy improved to the same extent. About 35 percent of adolescents getting sugar pills were rated as much improved.

Adolescents with suicidal thinking at the start of the trial showed declines in such behavior irrespective of whether they got medication, psychotherapy or sugar pills. Insel said this highlighted the importance of recognizing depression and treating it.

"Untreated depression is by itself fatal, and even placebo with usual care will confer some protection," he said.

Critics say the drugs make patients suicidal by causing violent inner restlessness, but proponents advocate alternate theories about why an antidepressant might increase the risk of suicide: Some patients might not have depression but manic-depressive, or bipolar, disorder, and an antidepressant could push one into mania.

Or healing depression might itself briefly increase the risk of suicide. "I was taught that as a resident in 1976," Insel said. "A person who is severely depressed is at greatest risk [for suicide] not at the deepest point of depression, but as they are coming out of it."

The study found that 11.9 percent of adolescents given Prozac harmed themselves or others. The category included suicide attempts, suicidal thinking, self-harm without suicidal thinking and aggressive behavior. And 8.4 percent of those who got a combination of Prozac and cognitive behavior therapy, whereas 4.5 percent of those who got just cognitive behavior therapy, showed such adverse effects. In the comparison group, adolescents getting a placebo, 5.4 percent showed such harmful behaviors.