Objectives. To determine the prevalence and meaning of the antiperinuclear factor (APF) and antikeratin antibodies (AKA) in a group of patients with polymyalgia rheumatica (PMR) and/or giant cell arteritis (GCA) and to evaluate the usefulness of APF and AKA in distinguishing early rheumatoid arthritis (RA) from PMR/GCA in everyday clinical practice. Methods. Eighty patients with PMR and/or GCA were compared with 44 patients older than 50 years and admitted during the same period for evaluation of a possible diagnosis of RA. All the patients underwent a standardized evaluation including a physical examination and diagnostic investigations. Follow-up was 2 years. Results. Fourteen (17.5%) patients in the PMR/GCA group had APF in titers ≥ 1/80, but only six (7.5%) had titers > 1/80. No significant differences were found between the subgroups with and without APF. None of these patients had detectable AKA. As compared to rheumatoid factors (RF), APF and AKA showed poorer performances (sensitivity, 60%; specificity, 91%; efficiency, 77%) for diagnosing RA versus PMR/GCA. Combining RF and APF resulted in increased specificity (96%). Conclusions. The prevalence of APF in titers ≥ 1/80 was high in PMR/GCA, but few patients (7.5%) had titers > 1/80. Presence of APF was not associated with specific features. All these patients were negative for AKA. RF has the best performance characteristics for diagnosing RA in older patients; RF combined with APF was more specific than RF alone.