The ability of 12 commercially available cardiotomy reservoirs to remove bubbles from aspirated blood was investigated by means of a simulated cardiopulmonary bypass circuit and an ultrasonic microbubble detector. Performance varied considerably. The number of gaseous microemboli remaining after passage of blood through the reservoir was reduced by (a) holding the blood in the reservoir, (b) reducing the volume of air mixed with the aspirated blood, and (c) using a reservoir that did not induce turbulence and that contained integral micropore filtration material. Further micropore filtration of the blood after passage through the cardiotomy reservoir was beneficial, and significantly more bubbles were extracted when the microfilter was sited below the reservoir than when it was placed in the arterial line.