Thirty patients with pauciarticular or polyarticular juvenile rheumatoid arthritis entered a randomized, 24-month, parallel trial of synovectomy (n = 15) versus no synovectomy (n = 15). The joints studied were 18 wrists, 8 ankles and 4 knees. The synovectomies were performed as radically as possible. Joint motion was slightly decreased in the operated joints, especially for passive movements during the first 6 months following surgery. Swelling and disease activity of the joints studied as well as subjective evaluation of joint pain improved in the synovectomy group compared with the non-synovectomy group. This improvement seemed to continue for at least 2 years.