Acute respiratory distress syndrome (ARDS) is conventionally treated by mechanical ventilation and administration of high O-2 concentrations, which often injure the patients' lungs. Physiological O-2 amounts may be effectively supplied by extracorporeal membrane oxygenation (ECMO) with minimal insult to lung tissue, improving survival of ARDS infants only. In the 1980s, apneic oxygenation and low frequency ventilation combined with extracorporeal CO2 removal (ECCO2R) from small vessels was reported to improve survival in the treatment of ARDS adults. In this treatment, membranes are only used to remove CO2 from Slowly flowing blood. Under these conditions, the use of hemodialysis hydrophilic membranes for ECCO2R could significantly improve the treatment biocompatibility. This paper reports an investigation aimed at analyzing in vitro the effect of the operating conditions on CO2 removal with hydrophilic polysulfone membranes from carbonated pig blood into an oxygen-rich gas stream. The CO2 removal rate and efficiency of clinical-size dialysis modules was investigated as a function of dissolved CO2 partial pressure (PCO2), blood and gas flow rate, and membrane surface area. Membrane CO2 removal rate increased with increasing PCO2, blood flow rates, and membrane surface areas approaching 18% of the CO2 physiological production rate. Data analysis with a state-of-the-art model suggests that liquid water permeation affects the module CO2 removal capacity, possibly by forming a stagnant liquid layer outside the membranes or by favoring channeling of gas flow. It was concluded that modules with hydrophilic membranes are feasible for ECCO2R at low blood flow rate. Their capacity may still be improved by minimizing liquid water permeation across the membranes, and by optimizing module design and operation to prevent gas channeling. (c) 2005 Elsevier B.V. All rights reserved.