To analyze the validity of baseline lung function parameters as predictors of maximal exercise ventilation (VEmax) in patients with chronic obstructive pulmonary disease (COPD), we studied 33 stable patients (FEV1 43.6 +/- 16.8%, FEV1/FVC% 48.4 +/- 9.2, FRC 156.8 +/- 32.7% and RV 212 +/- 53.9%). The sample was later divided into 3 groups based on severity of disease (severe, moderate or mild) in order to determine whether ability to predict VEmax increased with airways obstruction (FEV1 26.9 +/- 4.9%, 40.2 +/- 3.5% and 63.9 +/- 10%, respectively). The patients underwent lung function testing at rest and after a progressive stress test, with the maximal reading taken with the subject on a tread mill. We found greater correlation between VEmax and RV, FEV1 and FRC (r = -0.77, 0.75 and -0.74, respectively); the correlation was stronger in patients with severe COPD, in whom FEV1 was under 35% of the predicted value (r = -0.88, 0.753 and -0.83, respectively). Correlation decreased or disappeared with less functional impairment. Prediction of VEmax was more reliable with equations that employed FEV1 accompanied by data reflecting degree of insufflation, RV or FRC (VEmax = 45.2 +/- 8.98 x FEV1 - 5.07 x RV; r2 = 0.72) than with equations based on FEV1 alone (VEmax = 14.79 + 15.03 x FEV1; r2 = 0.56). We therefore conclude that ventilatory limitation during exercise in patients with COPD is better defined by considering parameters related to lung insufflation along with those reflecting degree of expiratory obstruction, given that the former affect the greater or lesser efficacy of muscles under stress.(ABSTRACT TRUNCATED AT 250 WORDS)