The aim of the study was to assess the prevalence of rheumatoid factor (RF) in a community-based rheumatoid arthritis (RA) series. The subjects of the series represented prevalent RA cases in the Kuusamo community in Northern Finland with 13,000 adult inhabitants. The patients were selected from the official Finnish data registers and from among the subjects who had consulted the local general practitioners due to rheumatic complaints in the recent years, using the American Rheumatism Association (ARA) 1987 classification criteria for inclusion. For this study, ninety-five out of the 103 RA patients so found were RF-tested by immunoturbidimetry. At the time of the study. 71 (75%) of the 95 cases were RF positive, 83 (87%) being 'ever' positive (in one case the early RF status was unclear). Our result contrasts with the much lower prevalence figures (25-60%) obtained from the earlier cross-sectional population-based RA studies, which have used the ARA 1958 definite RA as the inclusion criterion of the subjects. All the eleven patients with RF-negative RA had erosive joint disease. The RF-negative RA patients had a significantly lower frequency of HLA-DR4 (18%) than the RF-positive ones (58%), p < 0.05. We found a high frequency of RF among prevalent community-based RA cases meeting the ARA criteria. According to our results, RF-negative non-erosive RA is very rare among cases selected with the above methods.