Age and sex-adjusted first admission rates for operationally-defined schizophrenia and other non-affective psychosis in different ethnic groups were calculated over the period 1988-1992 in a defined catchment area in South London. Standardized rates for schizophrenia, corrected for age- and gender-related under-reporting in the 1991 census and a 20% underestimate of the size of the ethnic minority populations in the area, were not only higher in the Afro-Caribbean group (SMR: 3.1; 95% C1:2.0-4.7), but also in the African group (SMR: 4.2; 95% C1: 2.8-6.2). It was further found that higher rates were not specific to schizophrenia. These findings suggest that some common factor associated with ethnic minority membership is important in producing an excess of psychotic illness.