Relaxed expiratory volume-time profile has been frequently analysed by fitting exponential functions of time to one- or two-compartment models. In the latter case, the two exponential constants are assumed as representing the time constants of both compartments. Least-square fittings on the experimental data of five consecutive mechanically ventilated supine patients with acute respiratory distress syndrome (ARDS) were performed using rate-constants (flow/volume ratio) as parameters in order to obtain the model matching. Passive expiratory volume-time curves were recorded under PEEP = 0 and 13.6 +/- 3.3 S.D. cmH2O conditions. Model matching was optimal with significant, reliable parameter values. As a result, the use of a PEEP in ARDS patients: (a) delayed expiration; (b) decreased the percentage initial volume contribution of the slow-emptying compartment; and, (c) modified the interaction between compartments. The volume-time profile of the second compartment was found to increase at the beginning of expiration, and, then, progressively decayed towards zero, showing a maximum, although the overall curve decreased throughout expiration.