Given something of an international panic about people with a mental disorder, perhaps the most important contribution to rational practice is to emphasise that such disorders account for a minute proportion of society's violence. In countries with modest criminal homicide rates of, say, 2-6 per 100000 population, people with schizophrenia are overrepresented in criminal homicide statistics compared with their representation in the general population.1 In Denmark there is even some evidence of an increase, since 1959, in the number of men with schizophrenia who kill someone within their family.2 Criminal homicide, however, is a highly selected type of unnatural death even in peacetime.

For non-fatal violence, in the only true community survey of violence and mental disorder,3 people with schizophrenia, major affective disorder, substance misuse, or a combination of disorders were overrepresented in the violence groups. Even so, less than 10% of the people with pure schizophrenia reported having been violent in the 12 months before the interview. Furthermore, they accounted for less than 3% of the total violence reported in the sample. There are no data to support the media caricature of people with a mental disorder, the shunning of former patients by employers or neighbours, or the laws increasing constraints proposed by politicians pandering to public fears. If most of society can remain untroubled by schizophrenia, however, the 1% or so of the population that suffer from it, their families, and the victims of those few schizophrenic people who become violent deserve more than reassurances about probabilities.