Forty-six patients with classical or definite rheumatoid arthritis participated in a prospective clinical trial comparing auranofin 6 mg/day (26 patients) with D-penicillamine 500 mg/day (20 patients) during one year. NSAIDs were also given throughout the study period. After the first year, patients receiving auranofin with a satisfactory response continued for a further two years with a reduced dose of 3 mg/day. However the 6 mg dose could be reinstituted in patients showing deterioration after dose reduction. This paper only discusses the long-term treatment with auranofin. Seven out of 26 patients did not complete the one year treatment period; three because they did not return to follow-up, one because of inefficacy, and 3 because of untoward events. During the second year 5 more patients discontinued treatment, one because he was lost to follow-up, two because of inefficacy, one because of untoward events and another one because of a surgical procedure of the left knee. Two more patients discontinued auranofin treatment during the third year, one because of a flare up of his disease activity, and one because of a rash. Statistically significant improvements in the number of tender joints, activity and articular indices, duration of morning stiffness, pain score and ESR were observed at each time analysed. Statistically significant reductions in the number of swollen joints were seen throughout the first two years of treatment. Increases in grip strength were statistically significant at 6, 24 and 30 months. A statistically significant reduction was seen after 6 months of treatment in serum IgA and IgM concentrations.(ABSTRACT TRUNCATED AT 250 WORDS)