Fifty-three patients with rheumatoid arthritis who required immunosuppressive therapy were commenced in a randomized trial comparing azathioprine to weekly oral pulse methotrexate. After an initial 24-week period, both groups had significantly improved from baseline measures of pain and functional capacity and there were no significant differences in clinical outcomes between the two groups. Laboratory variables of disease activity showed a significant improvement in haemoglobin and ESR in the methotrexate group. Subsequently, the patients were followed for up to 3 years. After one year, more than half of the patients in both groups had discontinued therapy due to inefficacy or adverse events. Adverse effects were more frequent in the patients treated with methotrexate, but withdrawal rates were similar in both groups. From these data, the probability of a patient continuing therapy with either agent for greater than 18 months is low.