At present, air handling of a membrane oxygenator is generally studied by using an ultrasonic sound bubble counter. However, this is not a quantitative method and it does not give any information on where air was entrapped in the oxygenator and if it eventually was removed through the membrane for gas exchange. The study presented here gives a novel technique for the determination of the air-handling characteristics of a membrane oxygenator. The study aimed at defining not only the amount of air released by the oxygenator, but also the amount of air trapped within the oxygenator and/or removed through the gas exchange membrane. Two neonatal membrane oxygenators without the use of an arterial filter were investigated: the Polystan Microsafe and the Dideco Lilliput. Although the air trap function of both oxygenators when challenged with a bolus of air was similar, the Microsafe obtained this effect mainly by capturing the air in the heat exchanger compartment while the Lilliput did remove a large amount of air through the membrane. In conclusion, the difference in trap function was most striking during continuous infusion of air. Immediate contact with a microporous membrane, avoidance of high velocities within the oxygenator, pressure drop, transit time and construction of the fibre mat all contribute to the air-handling characteristics of a membrane oxygenator.