It has been shown that expiratory muscle contraction leads to an overestimation of intrinsic positive end-expiratory pressure (PEEPi). To quantify this overestimation, we compared PEEPi, measured during spontaneous breathing (SB) by the end-expiratory airway occlusion technique (PEEPi,occl) with static PEEPi (PEEPi,st). PEEPi,st was measured using end-expiratory airway occlusion during simulation of SB by the ventilator with the patient relaxed, and was considered to represent the "gold standard" for PEEPi,occl. Twelve ventilator-dependent patients were studied during SB (pressure support 5-7 cmH2O). Full mechanical ventilation was resumed when they were unable to sustain SB. Subsequently, by manipulating the variables of the ventilator, we simulated the pattern of SB and measured PEEPi,st, corresponding to PEEPi,occl. On the basis of the presence or absence of expiratory rise in gastric pressure (Pga) (rapid drop of end-expiratory Pga at the beginning of inspiration, Pga,exp,rise), and abdominal muscle electromyographic (EMG) activity, patients were subdivided into those either actively (Group 1) or passively expiring (Group 2). In Group 1 (8 patients), PEEPi,occl was higher than PEEPi,st (13.3+/-2.0 vs 6.8+/-1.1 cmH2O; p<0.01). PEEPi,occl-Pga,exp,rise (6.9+/-1.1 cmH2O) was quite similar to PEEPi,st; their mean difference was 0.03 cmH2O with limits of agreement -0.48 to +0.53 cmH2O. In Group 2, PEEPi,occl was similar to PEEPi,st. We conclude that, in actively expiring patients, an accurate estimation of the actual PEEPi,st can be obtained by subtracting Pga,exp,rise from PEEPi,occl.