The airway response to exercise and inhaled terbutaline was assessed in 25 patients with cystic fibrosis (CF), seeking evidence for the possible deleterious effects of bronchial muscle relaxation. We postulated that "early" and "late" flows, taken from the full maximum expiratory flow volume curve, might move paradoxically in patients with unstable airways. Oxygen saturation was measured continuously; desaturation occurred early in exercise with partial recovery thereafter. This was unrelated to changes in expiratory airflow measurements. Both during and after exercise, and after inhaled bronchodilator, changes in expiratory airflow measurements were strikingly variable. Changes in individual measurements should be interpreted in relationship to the within-subject variability of the test in patients with CF. During exercise, there was a significant increase in mean FEV1; this was most marked in patients with worst lung function. Two patients (both with severe lung disease) showed paradoxical changes in early and late flows. After exercise, only two patients showed the asthmatic pattern of postexercise bronchoconstriction. After inhaled bronchodilator, the group as a whole showed small but statistically significant increases in expiratory airflow measurements. Those with highest baseline FEV1 had the greatest bronchodilator response; this is the opposite of the pattern observed in asthma. Paradox did not occur after bronchodilators and only one patient showed a significant fall in late expiratory airflow. This pattern of expiratory airflow changes is compatible with the concept of airway instability in which any beneficial effects of bronchial tone reduction are canceled out by the effects of compression of damaged airways rendered more compliant by loss of bronchial wall tone. We did not observe any clinically important deleterious effects from this mechanism.