Thirty-five male patients, aged 34-79 yr, with definite rheumatoid arthritis (RA) were recruited from out-patient clinics and randomized to receive monthly injections of testosterone enanthate 250 mg or placebo as an adjunct therapy for 9 months. Endpoints included disease activity parameters and bone mineral density (BMD). At baseline, there were negative correlations between the ESR and serum testosterone (r = -0.42, P < 0.01) and BMD (hip, r = -0.65, P < 0.01). A total of 29.6% of all patients had at least one vertebral fracture, most having multiple fractures. Back pain, however, was not more prevalent in fracture patients (55% vs 50%). Disease activity was significantly higher in the fracture group (joint score P < 0.05, rheumatoid factor P < 0.01). Thirty patients completed the trial, 15 receiving testosterone and 15 receiving placebo. There were significant rises in serum testosterone, dihydrotestosterone and oestradiol in the treatment group. There was no significant effect of treatment on disease activity overall, five patients receiving testosterone underwent a "flare'. Differences in mean BMD following testosterone or placebo were non-significant (spine: +1.2% vs -1.1%; femur: -0.3% vs +0.3%). There was no suggestion of a positive effect of testosterone on disease activity in men with RA.