OBJECTIVE To determine the effects of a short course of a low dose of a glucocorticoid agent on bone mass. METHODS Double-blind, placebo-controlled, randomized study. METHODS Outpatient clinic of a university hospital. METHODS Forty patients with active rheumatoid arthritis. METHODS All patients started receiving intramuscular gold salts. In addition, they were randomly allocated to receive either prednisone or placebo. The initial dose was 10 mg/d, which was tapered between weeks 12 and 20. Thereafter, patients were followed for an additional 24 weeks. METHODS Lumbar bone mineral density was measured with dual-energy, quantitative computed tomography in a trabecular and a cortical region of interest. RESULTS Despite favorable effects on disease activity and functional capacity, trabecular bone mineral density decreased in the prednisone-treated patients between baseline and week 20 (mean change, -8.2%; 95% CI, -12.7% to -3.7%; P = 0.001). Little change was found in the placebo-treated patients (P > 0.2), and the prednisone group had a greater mean bone loss than the placebo group (9.5%; CI, 3.4% to 15.6%; P = 0.003). After discontinuation of prednisone, an increase was found in trabecular bone mineral density between weeks 20 and 44 (mean change, 5.3%; CI, 0.7% to 9.9%; P = 0.03). Little change was found after withdrawal of placebo (P > 0.2). The mean improvement in the prednisone group was 6.8% (CI, 0.8% to 12.8%; P = 0.03) greater than for placebo. In both treatment groups, cortical bone mineral density did not change markedly in either period (P > or = 0.2). CONCLUSIONS Low doses of glucocorticoid agents cause marked vertebral trabecular bone loss in the initial months of therapy in patients with active rheumatoid arthritis. After discontinuation of treatment, this bone loss seems to be (partially) reversible.