The results of a randomized synovectomy trial are reported and the history of early knee synovitis in 121 patients with rheumatoid arthritis is described. Conservative treatment (rest in splints, non-steroidal anti-inflammatory drugs and gold or an anti-malaria agent) for 4-5 months led to improvement in 81 of the 121 patients with early knee synovitis. In one-third of these 81 patients, the improvement continued, according to their own and the doctor's opinion, during 5 years of follow-up. Ten of the 22 patients who fulfilled the criteria for the synovectomy trial were operated upon. They showed improvement for 4 years. Over the 5-year period, two-thirds of the group were constantly seropositive or seronegative and in the other third the serology varied, but such that at each assessment 50% of the group were seropositive. Radiologically, all of the patients showed deterioration. The surgical group deteriorated more slowly for the first 2-3 years, but then showed the same pattern as the other groups. The disappearance of the clinical features of synovitis does not mean that radiological deterioration has ceased.