This study examined the effects of bronchodilator-induced reductions in lung hyperinflation on breathing pattern, ventilation and dyspnoea during exercise in chronic obstructive pulmonary disease (COPD). Quantitative tidal flow/volume loop analysis was used to evaluate abnormalities in dynamic ventilatory mechanics and their manipulation by a bronchodilator. In a randomised double-blind crossover study, 23 patients with COPD (mean+/-SEM forced expiratory volume in one second 42+/-3% of the predicted value) inhaled salmeterol 50 mug or placebo twice daily for 2 weeks each. After each treatment period, 2 h after dose, patients performed pulmonary function tests and symptom-limited cycle exercise at 75% of their maximal work-rate. After salmeterol versus placebo at rest, volume-corrected maximal expiratory flow rates increased by 175+/-52%, inspiratory capacity (IC) increased by 11+/-2% pred and functional residual capacity decreased by 11+/-3% pred. At a standardised time during exercise, salmeterol increased IC, tidal volume (V-T), mean inspiratory and expiratory flows, ventilation, oxygen uptake (V'O-2) and carbon dioxide output. Salmeterol increased peak exercise endurance, V'O-2 and ventilation by 58+/-19, 8+/-3 and 12+/-3%, respectively. Improvements in peak V'O-2 correlated best with increases in peak VT; increases in peak VT and resting IC were interrelated. The reduction in dyspnoea ratings at a standardised time correlated with the increased VT. Mechanical factors play an important role in shaping the ventilatory response to exercise in chronic obstructive pulmonary disease. Bronchodilator-induced lung deflation reduced mechanical restriction, increased ventilatory capacity and decreased respiratory discomfort, thereby increasing exercise endurance.