A 6-month double-blind, parallel, randomized, placebo-controlled multicenter trial of auranofin (0.15-0.20 mg/kg/day) was conducted in 231 children with juvenile rheumatoid arthritis (JRA) in the United States and in the Union of Soviet Socialist Republics. Approximately 80% of the children had polyarticular disease. The auranofin-treated patients showed greater mean decreases from baseline in 11 of the 12 articular disease indices measured than did the placebo-treated subjects after 3 months of therapy, and in 9 of the 12 indices after 6 months. However, the actual intergroup mean differences were relatively small and were not statistically significant. According to the physician's global assessment, 69% of the auranofin-treated patients and 61% of the placebo-treated patients demonstrated clinically significant improvement from baseline after 6 months (P = 0.24). Children whose disease onset occurred less than 2 years prior to entry improved more than did those who had arthritis for a longer period. In addition, those with polyarticular involvement at baseline improved more than did patients with mild disease. However, these relationships were observed in both the auranofin- and placebo-treated groups, and again, there were no significant intergroup differences. Diarrhea was the most common adverse effect of auranofin. We conclude that the clinical efficacy of auranofin is modestly higher than that of placebo in the treatment of JRA, as evidenced by the consistent trends observed in the data. However, the magnitude of the individual intergroup differences is not statistically significant. Auranofin appears to be very safe in children with JRA.