Sixty-eight patients with definite or classic rheumatoid arthritis were enrolled in a double-blind, controlled study of gold salt therapy. The initial phase compared weekly injections of gold thiomalate to placebo. One-third of the gold group were withdrawn from the trial because of toxicity and one-fourth of the controls because of no benefit. The gold-treated patients showed slight, but definite, improvement in all parameters measured, although only the change in sedimentation rate was statistically significant. In Phase 2, designed to ascertain the effects of maintenance therapy, the gold group showed no increase in the number of involved joints, improved their grip strength and had a fall in erythrocyte sedimentation rate. Over the same period the placebo group deteriorated in all these parameters. The number of patients treated, however, was too small to allow definite conclusions.