The aim of this study was to define the most useful index of expressing bronchodilator response and to distinguish between asthma and COPD. A prospective study was carried out of bronchodilator response in 142 asthmatics and 58 COPD patients in a university hospital. Reversibility was expressed as: 1. absolute change (delta abs); 2. % of initial (delta %init); 3. % of predicted (delta %pred) and 4. % of maximum possible response (delta %max). Dependence on forced expirations volume in 1 sec (FEV1) as % of predicted and sensitivity and specificity for diagnosis of asthma were established. A relationship between delta abs and initial FEV1 was not found in asthma (delta abs vs. % initial FEV1. r = 0.07) or COPD (r = 0.02). delta %pred did not show a correlation in asthma (r = 0.10) or COPD (r = 0.06). delta %init was dependent on the baseline value in asthma (r = 0.38, P < or = 0.001) but not in COPD (r = 0.18, P = n.s.). delta max was dependent in both. The combination of best sensitivity and specificity to separate asthma and COPD was obtained with delta abs (70.4 or 70.6%). The worst specificity for asthma diagnosis was obtained with delta %init (50%). The best likelihood ratios were obtained with delta abs and delta %pred and the worst likelihood ratio with delta %init. delta %init is not recommended as an index for differential diagnosis between asthma and COPD; 2) delta %init overscores bronchodilator response in patients with low FEV1. The independence of each bronchodilator response index should be verified in clinical trials for each selected sample.