In 1964 Nienhuis and Mandema reported the presence of antibodies against cytoplasmic granules in buccal mucosal cells in the serum of 50% of patients with rheumatoid arthritis (RA). Although they reported a good specificity for RA of these so-called antiperinuclear antibodies (APF), their results never threatened the monopoly of the rheumatoid factor as a serological tool for the diagnosis of RA. A re-evaluation with improved immunofluorescence methods showed a frequency of the APF of 78% in 103 patients with RA. The latex test and the Waaler-Rose test were positive in only 70% and 58% respectively of these patients. Only 15% of the RA patients were negative for all 3 tests. Thus, 40% of patients who were seronegative by the traditional methods gave a positive result on performance of the APF test. The high sensitivity of the APF test was combined with a good specificity, for the frequency in patients with other autoimmune diseases or degenerative joint disease and in healthy subjects was low. For the serodiagnosis of RA it seems best to combine the use of the APF test with one for rheumatoid factor.