To determine the effects of a step change in end-expiratory pressure on functional residual capacity (FRC) and lung-thorax compliance (CLT), 10 cm H2O positive end-expiratory pressure (PEEP) was applied in eight patients who needed mechanical ventilation for acute pulmonary failure. Of the total change in FRC, 66 +/- 5.3 per cent (mean +/- SEM) was complete within the next breath, and 90 per cent change was achieved in 4.6 +/- 1.4 breaths (24 +/- 6.4 sec). There was no statistically significant difference between times to 90 per cent FRC change with application and with removal of PEEP. In another 13 patients, PEEP was increased in 5 cm H2O steps from 3 to 18 cm H2O. Mean FRC at 3 cm H2O PEEP was 1.51 +/- 0.20 1 (55 +/- 7.0 per cent predicted supine value). Mean CLT did not change significantly until 18 cm H2O PEEP was reached, at which point it decreased (P < 0.005). The static compliance derived from change in FRC (deltaFRC/deltaPEEP) increased with increments of PEEP (P < 0.05) compared with the initial level. At PEEP levels of 8 and 13 cm H2O, mean FRC was larger than would be predicted from mean CLT (P < 0.005), but it was not significantly different at 3 cm H2O PEEP. The lung component accounted for 62 +/- 3.7 per cent of the lung-thorax compliance difference. These data define a time-dependent increase in lung volume that resembles pressure-volume hysteresis in normal man. Possible mechanisms include surface tension changes, recruitment of nonventilated lung, and stress relaxation of lung and chest wall. This study may explain the greater efficiency of PEEP compared with large tidal-volume ventilation in increasing PaO2 in patients with acute pulmonary failure.