Between 1968 and 1981 there were roughly equal numbers of male and female schizophrenic parasuicides in Edinburgh: males were significantly younger than females at parasuicide (33 v. 37 years), and had received a diagnosis of schizophrenia for a shorter time (3.4 v. 7.2 years). The clinical, epidemiological and social characteristics of such parasuicides are, by and large, those that might be expected from a knowledge of the epidemiology of schizophrenia. Eight per cent of a sample of Edinburgh schizophrenic first-ever parasuicides subsequently committed suicide, when followed up for up to 14 years. This percentage is close to what would be expected for later suicide in non-schizophrenic parasuicides. When all the suicides were considered, similar numbers of male and female schizophrenics killed themselves; their mean ages at death were 42 and 43 years respectively; and the mean duration of the schizophrenic illness was 10 years in both cases, according to hospital records. We found slight evidence that parasuicide is commoner in schizophrenic suicides than in schizophrenic controls but little evidence to support the contention that there may be a specific relationship between suicide (or parasuicide) and the presence of auditory hallucinations, recent discharge from in-patient care, the use of depot neuroleptic medication, or recent parasuicide. From the point of view of the prevention of suicide and parasuicide in schizophrenics, it is noteworthy that we found few differences between the clinical characteristics and management of schizophrenic parasuicides and suicides, and schizophrenic controls.