The benefits of chronic systemic corticosteroids for patients with chronic obstructive pulmonary disease (COPD) are not well established. To determine whether chronic corticosteroid treatment can be safely withdrawn in "steroid-dependent""COPD patients, we performed a double-blind, placebo-controlled study of 38 patients with steroid-dependent COPD. Patients were randomly assigned to receive their usual maintenance prednisone dose for 6 mo (continuous group) or to be withdrawn from prednisone at a rate of 5 mg per week (demand group). The number of COPD exacerbations per patient (primary outcome) was 2.5 +/- 2.7 (mean +/- SD) in the continuous group and 2.7 +/- 2.5 in the demand group (p = 0.60, 95% confidence interval for the difference: -1.1 to 1.7). Spirometric results, dyspnea, and health-related quality of life did not differ significantly in the two groups. The average daily corticosteroid dose was 10.7 +/- 5.2 mg in the continuous group and 6.3 +/- 6.4 mg in the demand group (p = 0.003). Weight decreased in the demand group by 4.8 +/- 2.0 kg, compared with an increase in the continuous group of 0.5 +/- 3.5 kg (p = 0.007). Discontinuation of chronic systemic corticosteroid treatment in steroid-dependent COPD patients did not cause a significant increase in COPD exacerbations, but did reduce total systemic corticosteroid use and body weight. Larger studies may be warranted to establish the relative risks and benefits of chronic corticosteroid treatment of patients with COPD.