Synovectomy of the knee results in a satisfactory degree of pain relief in a patient with rheumatoid arthritis. Pain relief was equally good in our series in knees with clinically active synovitis, regardless of the amount of joint destruction, provided no gross malalignment, instability, or degeneration of joint surfaces was present. In 64 per cent of the knees there was some loss of range of motion; in 13 per cent, a gain in range of motion; and in 23 per cent, no change in the total range of motion. Only two knees in the entire series underwent fibrous ankylosis. The patients' estimates of their disease activity after surgery correlated very closely with the pain relief obtained. Most of those who estimated their activity to be very low had good pain relief, while those who had more active disease had less likelihood of a satisfactory result. Although long-term satisfactory results seemed to indicate that synovitis had been effectively suppressed by surgical synovectomy, the data in this series do not prove that synovectomy prevents destruction in the rheumatoid knee.