The authors investigated the association between schizophrenia and rheumatoid arthritis. The design is a population-based case-control and follow-up study. The cases were 20495 patients admitted for schizophrenia and registered in the Danish Psychiatric Case Register. A total of 204912 persons matched on age and gender and chosen from the general population served as controls. Admissions for rheumatoid arthritis and other non-autoimmune, musculoskeletal disorders were checked in the Danish National Patient Register. Odds ratios and relative risks were estimated by the Mantel-Haenszel estimator and Poisson regression. The same analyses were carried out for 10242 patients with bipolar affective disorder and 102420 controls for comparison. Individuals with schizophrenia had a reduced risk for being admitted with rheumatoid arthritis [odds ratio 0.44 (CI 0.24-0.81)] in the case-control study. A similar result was found in the follow-up study, but the incidence of the degenerative disorders in the musculoskeletal system was equally significantly lower in both studies. The incidence of rheumatoid arthritis among the bipolar patients was the same as in the control population. The negative association between schizophrenia and rheumatoid arthritis may thus be the result of ascertainment bias and selection due to under reporting and treatment of the medical illness. Clinicians are reminded of the difficulties in detecting medical illness among individuals with schizophrenia.