Pressure support (PS), a new mode of ventilatory assistance, is known to induce respiratory muscle relaxation. It was used to obtain reliable measurements of the compliance of the respiratory system (Crs) in awake subjects. PS was applied, through a mouthpiece, at four successive levels (0, 0.75, 1 and 1.25 kPa) to 30 healthy subjects. At the highest PS level, the subject's relaxation was obtained as assessed by a decrease in the occlusion pressure from 0.10 +/- 0.06 to 0.05 +/- 0.04 kPa, whereas the minute ventilation increased (from 7.5 +/- 1.5 to 13.8 +/- 3.3 l.min-1), and the end-tidal carbon dioxide tension (PCO2) decreased (from 5.0 +/- 0.4 to 3.2 +/- 0.5 kPa) below its apnoea threshold. In three subjects, respiratory muscle relaxation was confirmed by a fall in diaphragmatic electromyographic activity. Crs was calculated as the ratio of the tidal volume to the corresponding end-inspiratory airway pressure (i.e. PS level) since, at end inspiration, a zero-flow period was obtained. Crs was highly correlated (r = 0.77) to the height (Ht) of the subjects: Crs (l.kPa-1) = 3.56 x Ht (m) -4.86 (+/- 0.23), allowing normal values to be determined. In order to evaluate the applicability of the method to patients, Crs was measured in four patients with scoliosis, and was found to range from 45-82% of the predicted values. It is suggested that this simple method of Crs determination may be used to characterize various chest wall or pulmonary diseases.