Evidence on the safety and efficacy of methotrexate as a second- or third-line agent for treating patients with rheumatoid arthritis is reviewed. Four placebo-controlled clinical trials have documented short-term benefit from methotrexate; although true remission is rare, patients receiving methotrexate showed a 26% (95% confidence interval [CI], 17% to 35%) greater improvement in their inflamed joint count and a 39% (95% CI, 26% to 51.5%) greater improvement in pain than did controls receiving nonsteroidal anti-inflammatory agents with or without prednisone. With respect to long-term benefit, improvement usually occurs within 1 month, reaching a maximum at 6 and then leveling off for the duration of treatment; in some patients, the benefit may wane after an initial satisfactory response in the first 4 to 6 months. In one third of those given methotrexate, treatment had to be discontinued because of adverse effects, less than 1% of which were life threatening. Careful baseline and follow-up monitoring is recommended until more data on the safety of methotrexate are available.