Twenty-five patients in acute respiratory failure were randomised to receive either 5 cm of positive end-expiratory pressure (P.E.E.P.) or no-P.E.E.P. while weaning from controlled ventilation. The use of P.E.E.P. resulted in a significant reduction in the increase in alveolar-arterial oxygen tension gradient (AaDO21) which occurred in the group of patients who were converted from controlled ventilation to spontaneous ventilation without P.E.E.P. Patients who weaned without P.E.E.P. had a mean increase in AaDO21 of 102+/-35 mm Hg S.E. while those who weaned with P.E.E.P. had a mean increase of only 10+/-22 mm Hg (P less than 0-03). The use of P.E.E.P. was also associated with a significant improvement in the vital capacity and the maximum inspiratory force. Patients who weaned with P.E.E.P. had an increase in vital capacity of 258+/-108 ml (P less than 0-05) and an increase in inspiratory force of -15+/-5 cm H2O (P less than 0-01), while patients who weaned without P.E.E.P. did not have significant changes in these measurements. The use of P.E.E.P. during weaning may be helpful in patients who fail to wean because of the development of hypoxaemia due to rapid alveolar collapse, since P.E.E.P. appears to minimise the increase in intrapulmonary right-to-left shunt which normally occurs during weaning from controlled ventilation.