A longitudinal epidemiologic study has been conducted to estimate the incidence and prevalence of rheumatoid arthritis in an American Indian population, the Pima and Papago Indians of Arizona. Clinical, serologic, and radiologic data were collected during biennial examinations of subjects aged 20 years or more during the period 1967-1986. Rheumatoid arthritis was diagnosed by criteria for the active and the inactive disease. Age-adjusted to the 1980 US population at least 20 years of age, the prevalence of classical and definite rheumatoid arthritis in 1984 was 5.3% (3.23% in males and 6.95% in females), a rate appreciably higher than that reported in studies in Rochester, Minnesota, and in Hiroshima and Nagasaki, Japan. Among Pimas, during the study period, 70 incident cases of rheumatoid arthritis occurred. The age-adjusted incidence rate was 42.2 cases per 10,000 person-years (29.7 in males and 51.8 in females), 10.3 times as high as the age-adjusted rate in Rochester (4.1/10,000 person-years), and 5.7 times as high as in Japan (7.4/10,000 person-years). Rates generally increased with age. No secular trend was found. On the basis of both prevalence and incidence data, this study confirms that rheumatoid arthritis does not have uniform occurrence in different populations. This has to be taken into account in the search for the factors related to the differences in risk of disease.