Controlled mechanical ventilation with positive end-expiratory pressure(PEEP) is a widely accepted method for the treatment of acute respiratory failure. But artificial ventilation with large tidal voume, high airway pressure and an inspired oxygen concentration of l00% are necessary for adequate arterial oxygenation in severe acute pulmonary parenchymal failure. It has been suspected that such therapy may cause irreversible pulmonary damage. The extracorporeal respiratory support has been called extracorporeal membrane oxygenation (ECMO), extracorporeal CO, removal(ECCO2R) or extracorporeal lung assist(ECLA). They are prolonged venoarterial(VA) or venovenous(VV) bypass for the refractory hypoxemia which provides the diseased lung with rest. The respiratory rate can be lowered down to about 4~8 rates/min during ECLA. PEEP is known to improve the ventilation perfusion distribution with alveolar recruitment. Low frequency positive pressure ventilation is accompanied by PEEP to keep alveoli open during ECLA. The level of PEEP must be determined adequately because high PEEP may decrease the cardiac output especially during VV ECLA. The experimental model for acute respiratory failure was induced by the intravenous injection of oleic acid 0.07ml/kg on six mongrel dogs(15.8±0.7kg). After values of arterial oxygen saturation(SaO2) and end-tidal CO2(E(T)CO2) were stabilized on the monitor, we measured control values of hemodynamic parameters. The stepwise increase of PEEP from 5 cmH2O to 15 cmH2O via 10 cmH2O was followed by the comparison of the respective hemodynamic values at each PEEP with control. Compared with control, PEEP of 5 cmH2O did not cause any changes and PEEP of 15 cmH, decreased cardiac output. At PEEP of 10 cmHO, the cardiac output and oxygen flux was maintained with normoxia. It was concluded that PEEP of 10 cmHO can be applied to clinical VV ECLA.