Although recent guidelines for managing chronic obstructive pulmonary disease (COPD) recommend a trial of oral corticosteroids in the initial assessment, its prognostic value remains unclear. We prospectively studied 127 adults (64% men) with stable COPD (FEV1/FVC < 60%) over 1 year. At entry, we measured lung volumes, gas transfer factor, respiratory symptoms (by questionnaire), and peripheral blood eosinophil count. Skin-prick testing was done, and spirometry after nebulized 5 mg salbutamol and, after 2 weeks, oral prednisolone. Physician A gave all patients inhaled beclomethasone dipropionate (800 mcg/day), whereas physician B prescribed this only to those with a positive oral corticosteroid trial. At 1 year, spirometry and respiratory questionnaire were repeated, with an estimate of overall symptom severity on a visual analogue scale. Follow-up data were available in 104 (82%) patients. Of these, 32 (31%) were unresponsive to salbutamol and prednisolone; 48 (46%) were responsive to beta agonists but not to corticosteroids, and 24 (23%) responded to corticosteroids and salbutamol. Patients in all groups were comparable, except that the prednisolone responders had a higher mean eosinophil count (p < 0.001) and more were ex-smokers (p < 0.001). Only the response to oral prednisolone correlated with the change in prebronchodilator FEV1 over 1 year. Oral prednisolone responders had higher FEV1 at 1 year (p < 0.02) and significantly lower symptom scores (p < 0.02). In COPD, corticosteroid trials contribute information additional to that gained from nebulized bronchodilator reversibility testing. Patients with a positive response to a corticosteroid trial are more likely to have improved symptomatically and spirometrically at 1 year.