Pulmonary arterial occlusion pressure (Ppao) may not accurately reflect left ventricular filling pressure (LVFP) during ventilation with positive end-expiratory pressure (PEEP) because increases in pleural pressure (Ppl) increase measured intrathoracic vascular and cardiac chamber pressures relative to atmospheric while decreasing LVFP by decreasing venous return. Steady-state values of Ppao off PEEP are not useful in understanding the hemodynamics on PEEP because of changes in blood volume distribution and ventricular afterload associated with the removal of PEEP. We hypothesized that changes in Ppao immediately after abrupt airway disconnection from PEEP selectively reflect removal of PEEP-associated Ppl changes prior to other physiological changes. In pentobarbital-anesthetized closed-chest dogs, we compared absolute left atrial pressure and left atrial pressure relative to Ppl (Platm) prior to abrupt disconnection from 15 cmH2O PEEP (on-PEEP), with Ppao at its nadir (nadir Ppao) following airway disconnection in three sequential conditions: control (normal), after propranolol-induced acute ventricular failure, and after oleic acid-induced lung injury. For all conditions at low LVFP (less than 9 Torr), nadir Ppao reflects Platm better than on-PEEP Ppao, while at higher LVFP (greater than 9 Torr), on-PEEP Ppao better reflects Platm than nadir Ppao (P less than 0.05). Accurate predictions of on-PEEP Platm can be made using both on-PEEP and nadir Ppao values in a multiple regression equation.