The pattern of lung emptying was studied in 10 mechanically-ventilated patients with acute respiratory distress syndrome.At four levels of positive end-expiratory pressure (PEEP) (0, 5, 10 and 15 cmH 2 O) tracheal (Ptr) and airway pressures (Paw), flow (V9) and volume (V) were continuously recorded. Tidal volume was set between 0.5-0.6 L and V9/V curves during passive expiration were obtained. Expired volume was divided into five equal volume slices and the time constant (te) and effective deflation compliance (Crs eff ) of each slice was calculated by regression analysis of V/V9 and postocclusion V/Ptr relationships, respectively. In each slice, the presence or absence of flow limitation was examined by comparing V9/V curves with and without decreasing Paw. For a given slice, total expiratory resistance (Rtot) (consisting of the respiratory system (Rrs), endotracheal tube (Rtube) and ventilator circuit (Rvent)) was calculated as the te/Crs eff ratio. In the absence of flow limitation Rrs was obtained by subtracting Rtube and Rvent from Rtot, while in the presence of flow limitation Rrs equaled Rtot. The te of the pure respiratory system (ters) was calculated as the product of Rrs and Crs eff .At zero PEEP, ters increased significantly towards the end of expiration (52¡31%) due to a significant increase in Rrs (46¡36%). Application of PEEP significantly decreased Rrs at the end of expiration and resulted in a faster and relatively constant rate of lung emptying.In conclusion, without positive end-expiratory pressure the respiratory system in patients with acute respiratory distress syndrome deflates with a rate that progressively decreases, due to a considerable increase in expiratory resistance at low lung volumes. Application of positive end-expiratory pressure decreases the expiratory resistance, probably by preventing airway closure, and as a result modifies the pattern of lung emptying.