A committee chaired by Marion Ropes was appointed by the ARA to define criteria best suited to define RA. A number of physicians ‘particularly interested in rheumatic diseases in various sections of the United States and Canada’were asked to each contribute data from their clinics of the five most recent patients with definite RA, five recent patients with probable RA and five patients with no evidence of RA. In all 332 patients were analysed for presence of 11 criteria that had been selected on the basis of clinical experience by the authors. Sensitivity and specificity was calculated for presence of five or more criteria (definite disease), 3–4 criteria (probable disease), and less than 3 of the 11 criteria (no disease). In order to avoid inclusion of false positives, 19 exclusions were also listed.