We conducted a double‐blind, randomized trial comparing azathioprine (AZA) and methotrexate (MTX) in the treatment of patients with rheumatoid arthritis in whom parenteral gold and/or D‐penicillamine treatment had been unsuccessful. Patients were randomly assigned to receive either AZA (100 mg daily) or oral MTX (7.5 mg weekly). After 8 weeks, the dosage was increased depending on the clinical improvement. Sixty‐four patients were followed up for 48 weeks (33 AZA, 31 MTX). Comparison of values at week 24 with baseline values revealed significant improvement in 12 of 13 disease variables in the MTX group and in 6 of 13 in the AZA group. Comparison between the 2 treatment groups at 24 weeks, by area‐under‐the‐curve analysis, showed significantly more improvement in the MTX group in terms of the swollen joint count, pain score, erythrocyte sedimentation rate, C‐reactive protein level, hemoglobin level, thrombocyte level, and disease activity score. A significant overall clinical improvement (disease activity score) was found in 7 of 20 patients treated with AZA and 18 of 30 patients treated with MTX after 24 weeks of therapy, and in 6 of 12 AZA‐treated patients and 19 of 25 MTX‐treated patients after 48 weeks. The number of withdrawals due to side effects was significantly higher in the AZA group. After 48 weeks, only 12 patients from the AZA group (36%), but 25 from the MTX group (81%), were still using the initial drug. These results demonstrate MTX to be superior to AZA in the treatment of rheumatoid arthritis, with a more rapid clinical improvement which is sustained after 1 year, accompanied by a lower rate of serious adverse reactions.