We compared the effect of inhaled budesonide with placebo on decline in lung function and respiratory symptoms in a three-year study of patients with chronic obstructive pulmonary disease (COPD). We used a parallel-group, randomized, double-blind, placebo-controlled design, nested in an ongoing epidemiological survey. Patients were non-asthmatic subjects with a decreased ratio between forced expiratory volume in one second (FEV1) and vital capacity (VC); i.e., FEV1/VC < or = 0.7. All included patients had an FEV1 which was irreversible to both inhaled terbutaline and prednisolone. Two hundred and ninety patients were randomized to receive either budesonide, 1200 mcg. daily for six months followed by 800 mcg. daily for 30 months, or placebo for 36 months. Patients had a mean age of 59 years and their mean FEV1 was 2.37 liters or 86% of predicted. Crude FEV1 declines were 41.8 ml/year in the placebo group and 45.1 ml/year in the budesonide group. Using a regression model in the intention-to-treat population, patients in the placebo group had an FEV1 decline of 49.1 ml/year in contrast to 46.0 ml/year in the budesonide group; the estimated difference 3.1 ml/year (95% confidence interval--12.8-19.0) was statistically insignificant, p = 0.70. No effect of inhaled budesonide was seen on respiratory symptoms or number of exacerbations. These findings question the role of longterm inhaled corticosteroids in the treatment of mild-moderate COPD.