Background: A limited or absent bronchodilator response is used to classify chronic obstructive pulmonary disease (COPD) and can determine the treatment offered. The reliability of the recommended response criteria and their relationship to disease progression has not been established. Methods: 660 patients meeting European Respiratory Society (ERS) diagnostic criteria for irreversible COPD were studied. Spirometric parameters were measured on three occasions before and after salbutamol and ipratropium bromide sequentially or in combination over 2 months. Responses were classified using the American Thoracic Society/GOLD (ATS) and ERS criteria. Patients were followed for 3 years with post-bronchodilator FEV1 and exacerbation history recorded 3 monthly and health status 6 monthly. Results: FEV1 increased significantly with each bronchodilator, a response that was normally distributed. Mean post-bronchodilator FEV1 was reproducible between visits (intraclass correlation 0.93). The absolute change in FEV1 was independent of the pre-bronchodilator value but the percentage change correlated with pre-bronchodilator FEV1 (r=–0.44; p<0.0001). Using ATS criteria, 52.1% of patients changed responder status between visits compared with 38.2% using ERS criteria. Smoking status, atopy, and withdrawing inhaled corticosteroids were unrelated to bronchodilator response, as was the rate of decline in FEV1, decline in health status, and exacerbation rate. Conclusion: In moderate to severe COPD bronchodilator responsiveness is a continuous variable. Classifying patients as “responders” and “non-responders” can be misleading and does not predict disease progression.