The aim of the study was to determine the incidence of hospital care with the diagnosis rheumatoid arthritis (RA) among patients with schizophrenia, affective psychosis and neurosis compared with that among hospitalized patients in general. By means of the in-patient register of Stockholm County, a cohort was formed comprising all patients discharged with the diagnoses schizophrenia, affective psychosis and neurosis in Stockholm County during 1971, and a sample of all patients discharged for any diagnosis during the same year. We followed the groups in the in-patient register through 1981 in order to identify hospital episodes with the diagnosis RA. Observed and expected incidences of RA in hospital care were obtained using all hospitalized patients as a reference group. For schizophrenia and affective psychosis the incidence of RA was around half the expected, whereas for neurosis it was close to the expected incidence. With reservation for small numbers of observed cases, the results support the hypothesis of a reduced incidence of RA among patients with schizophrenia. The finding regarding affective psychosis was based on a smaller number of cases and merits further investigations.