To determine if long-term methotrexate-induced improvement of rheumatoid arthritis is sustained after the drug is discontinued, 10 unselected patients with responses to weekly oral methotrexate given for at least 36 months (mean 40.1) were randomly assigned to receive methotrexate or identical-appearing placebo tablets for two months. After one month, all five patients receiving placebo had to have the study terminated due to a flare of their disease manifested by statistically significant deterioration in multiple clinical parameters. It is concluded that patients receiving long-term methotrexate must continue the drug to maintain clinical benefits.