The results of a randomised trial of irradiation of the knee (synoviorthése), by intra-articular injection of yttrium-90, and surgical synovectomy have been compared in twenty knees in seventeen patients. The mean length of follow-up was 2 years. Relapse occurred in three out of ten irradiated knees, and in 2 out of 10 operated knees. Fewer irradiated knees were involved when generalised exacerbations of polyarthritis occurred. Irradiation carries a lower risk of complications than does synovectomy, it is more acceptable to patients, requires fewer days in hospital, and is cheaper; it would seem to be the treatment of choice in the older patient.