A comparison of placebo, auranofin, and parenteral gold sodium thiomalate therapy in 209 patients with active rheumatoid arthritis was performed in a 21-week prospective, controlled, double-blind multicenter trial. When the 161 patients who completed at least 20 weeks of treatment were analyzed for different degrees of response, no remissions were identified. When 50 percent or greater improvement of pain/tenderness scores were compared for end of trial versus entry values, 9 percent of placebo-treated patients, 34 percent of auranofin-treated patients, and 48 percent of gold sodium thiomalate-treated patients showed important improvement that was statistically significant for both gold treatments. When 50 percent improvement for joint swelling was analyzed, 12 percent of the placebo-treated group, 28 percent in the auranofin-treated group, and 37 percent in the gold sodium thiomalate-treated group showed this degree of improvement. Auranofin almost achieved statistical significance for improvement in joint swelling when compared with placebo (p = 0.07), but gold sodium thiomalate was much better than placebo (p = 0.009). There was no statistically significant difference between the two gold treatments. Thus it appears that a subset of patients had an important response to gold therapy that would not be evident by the usual analyses of mean or median changes. Analysis for predictors of response did not discriminate between responders and nonresponders. Because the trial was limited to 21 weeks of therapy, no prediction of the longer-term-effects, especially for auranofin, should be inferred.