A case-control study was performed using 90% of all first-contact patients with a clinical diagnosis of schizophrenia residing in the London borough of Camberwell between 1965 and 1984. Cases and controls were obtained from the Camberwell psychiatric case register. Controls were those presenting with first episodes of non-psychotic disorders, matched for age, sex and period. The risk of schizophrenia was greater in those of Afro-Caribbean ethnicity, irrespective of age, gender or place of birth. This risk increased over the study period. The results cannot be explained by changes in the age, gender or ethnic structure of the local population. Effects of misdiagnosis or change in diagnostic practice were reduced by using uniform operational criteria. Possible explanations include maternal exposure to unfamiliar infective agents, a differential fall in the age at onset of illness, or worsening social adversity.