We present a population-based, longitudinal study of all incident cases (N = 538) of schizophrenia in the London Borough of Camberwell between 1964 and 1984. Cases were selected from the Camberwell Cumulative Psychiatric Case Register. Case-notes were obtained, and further classified using a computerized operational check list for rating psychotic illness. Cases are not restricted to hospital discharges, as in previous studies, and account is taken of time at liberty to offend. In order to test the hypothesis that schizophrenia makes an independent contribution to criminality over other mental disorders, controls were chosen to be representative of nonschizophrenic mental disorders matched for age, sex and period. The rate of conviction is increased in women with schizophrenia compared to other mental disorders for most offence categories (rate ratio = 3.3). In men overall rates do not differ (rate ratio = 1.03), although there is an interaction between gender, schizophrenia and ethnicity, with young black men with schizophrenia being most at risk. The rate ratio for violent offences in men with schizophrenia is 3.8, confirming recent studies from Sweden. Subjects with schizophrenia were more likely to acquire any criminal record than those with other mental disorders. The rate of lifetime conviction was greater in those with schizophrenia than either a sample of working-class boys from the same area followed by Farrington & West, or National data. The risk of first conviction is increased by schizophrenia, unemployment, ethnic group, substance abuse and low social class, and decreased by being employed, married, female and older age of onset. Adjustment using survival analysis showed that schizophrenia made a small independent contribution to the risk of acquiring a criminal record (hazard ratio = 1.4) but gender, substance abuse, ethnicity and age of onset were more substantial. Previous criminality was the strongest independent association of post-illness conviction, with schizophrenia only a trend. Although subjects with schizophrenia were more likely to acquire a criminal record, criminal careers began later and were shorter than those of the controls. The risk of criminality increased throughout the study period, but suggestions of a specific increase in those with schizophrenia as a result of changes in community care were not confirmed. These results confirm that women with schizophrenia are at increased risk of acquiring a criminal record, but the effect in men is for violent convictions only. The strongest associations of criminal conviction remain those recognized in non-schizophrenic subjects.