Some years back, I was the attending psychiatrist in the emergency room when a young woman ran in, pursued by her boyfriend.

They had been arguing on the street when he became violent, and she desperately wanted protection.

Once inside, he began punching her, and the security guards immediately restrained him.

The boyfriend then demanded to see a psychiatrist, saying he was depressed. It did not take long to see that he was a con artist who was looking for a psychiatric label to excuse his violent behavior. The police were called, and he was arrested.

As the man was led out in handcuffs, he was barely able to suppress a mocking smile, underscoring how he had cynically exploited a common stereotype about the tendency of the mentally ill to be violent.

For decades, if not centuries, there has long been a lurid fascination with mental illness in popular culture, spurred in recent years by the unforgettable Norman Bates, the psychotic killer of Hitchcock's "Psycho" and Dr. Hannibal Lecter, the cannibalistic psychiatrist of "The Silence of the Lambs" and its sequel.

Not just the psychiatric patients, but the therapists themselves have also been widely depicted as deranged, if not downright homicidal.

Hollywood will be disappointed to hear it, but the real story about violence has just a minor role for the mentally ill.

The fact is that the contribution of psychiatric illness to all violence is quite small. For example, a Justice Department study found that people with histories of mental illness were responsible for 4.3 percent of all homicides in the United States in a year.

Even so, Americans remain anxious about the violent risk posed by the psychiatrically ill. So how large, really, is the danger?

Until recently, most of the studies that supposedly showed links between mental illness and violence focused on the rates of mental disorders among the violent or the rates of violence among psychiatric patients in hospitals.

Those studies were unpersuasive because of selection biases. Subjects who are arrested, incarcerated or hospitalized are by definition more likely to be violent or very ill and thus are not representative of the psychiatrically ill patients in the general population.

Newer studies have examined community samples and can give more accurate estimates of the risk of violence among the mentally ill.

Using the figures from the large Epidemiologic Catchment Area study sponsored by the National Institutes of Health, Dr. Jeffrey Swanson at the Duke University Medical Center reported that the incidence of violence was five times as high among people with serious mental disorders as it was among those with no psychiatric illness.

Among the serious disorders are schizophrenia, major depression and bipolar disorder. People who use alcohol or other drugs were found to be 12 to 16 times as likely to be violent as nonusers.

Because psychiatric patients have much higher rates of alcohol and substance abuse than the population as a whole, a crucial question concerns the extent that alcohol and drugs contribute to violent acts committed by the patients.

When alcohol was controlled for in the epidemiological study, the one-year prevalence of violence among subjects with major mental disorders dropped, to 7 percent from 20 percent.

That is still higher than the 2 percent prevalence of violence in subjects with no mental disorders or alcohol abuse.

A study published in The Archives of General Psychiatry by Dr. Henry Steadman of the University of Pittsburgh School of Medicine also reported that substance abuse was a strong risk factor for violence. Psychiatric subjects who did not abuse drugs were no more violent than members of a control group who had no diagnosed mental disorders and did not abuse drugs.

That study also followed the subjects for a year after they left their hospitals. It found that the highest rates of violence occurred before hospitalization.

But in the year of treatment after being discharged, when psychiatric symptoms are likely to be lower, the risk for violence among patients was no different from that of people who did not have psychiatric disorders.

This suggests that the violence is related to the level of psychiatric symptoms, not the psychiatric diagnosis itself. Though the study did not specifically monitor the treatments, it is reasonable to hypothesize that treating psychiatric illness may not just make patients feel better. It may also drastically lower the risk of violent behavior.

One surprising finding in this study was that when psychiatric patients were violent, they tended to be violent with family members or friends. Only 14 percent of all reported violence was directed at strangers.

It seems that a person is far more likely to be harmed by a psychiatrically ill family member or a friend than by a mentally ill stranger.

This is not to say that certain psychiatric illnesses like schizophrenia and bipolar disorder are not associated with an increased risk of violence. They certainly are, but the risk is only modest.

The fact is that an intoxicated alcoholic is far more dangerous than a schizophrenic person who is wandering the streets.

Why then does the public fear of the mentally ill so far exceed empirical reality? One factor is the news media coverage of the acts of the criminally insane. The public was transfixed by the horrific crimes of Jeffrey Dahmer, which, despite their rarity, burned the image of a deranged killer in the public mind.

A deeper clue to the stereotype of the violent psychiatric patient comes from Alfred Hitchcock.

What made his vision so terrifying was its accuracy. Ordinary people in the grip of ordinary passions are capable of extraordinary acts of horror and cruelty.

He was rumored to have said once, when traveling in the countryside, that it all seems so peaceful, but that the perfect crime happens there every day. Nothing, in other words, is more frightening than discovering the darkness beneath the apparently normal and mundane.

So it seems that the stereotype of the violent mental patient makes people feel safer by displacing and limiting the threat of violence to a small, defined group. After all, who could live with the reality that the nice guy next door might be a rapist or serial killer?