The efficacy, toxicity and possible steroid-sparing properties of auranofin in the treatment of elderly-onset rheumatoid arthritis (EORA) were studied in a 2 yr prospective double-blind placebo-controlled clinical trial. Sixty-five patients with onset of arthritis after the age of 60 yr were randomized to either auranofin 3 mg b.i.d. [n = 31, age 70 (61-84) yr, median (range)] or placebo tablets [n = 34, age 72 (60-81) yr]. Oral prednisolone, starting dose 7.5 or 20 mg daily, was used as a rescue drug in patients with intolerable joint pain and stiffness and with C-reactive protein (CRP) > or = 20 mg/l, and was tapered down according to protocol guidelines. Patients receiving auranofin continued therapy for a longer period of time (55% completers) than those on placebo medication (18% completers). The auranofin group consumed significantly less prednisolone, 2.64 (0-11.85) mg/day [median (range)], compared to 5.0 (0-18.33) mg/day in the placebo group (P = 0.006). No group differences at 2 yr follow-up were found for changes in joint pain (P = 0.49), number of swollen joints (P = 0.61), Health Assessment Questionnaire score (P = 0.18) and radiographic damage score (Larsen-Dale index) of the hands (P = 0.84). Within-group changes in radiographic scores were also insignificant. The drop-out rate due to adverse events was surprisingly higher in the placebo group (41%) than in the auranofin group (10%) and, as expected, higher due to lack of effect (29 and 16%). The results indicate that auranofin is safe, superior to placebo and has steroid-sparing capacity in the treatment of EORA. The favourable radiographic outcome in both groups needs confirmation in future studies.