Retrograde blood from the aortic cannula into the cardiopulmonary circuit may lead to aortic air emboli when nonocclusive centrifugal pumps are used. The authors tested a nonregurgitant, unidirectional valve containing a Teflon ball occluder to prevent backflow. In vitro measurements of leakage rate, forward flow pressure drop, burst strength, and hemolysis levels along with animal (n = 12) and human (n = 12) in vivo hematologic and hemolysis levels were measured. Data were analyzed by paired and unpaired Student's t-test. Pressure drop differences at flows of 5 l/min were 7.3 +/- 0.3 mmHg before and 7.6 +/- 0.1 mmHg after 10,800 cycles of pulsatile pumping. (P = NS). Leakage rate during this period at pressures of 100 mmHg was not significant. Volume required to close the Teflon ball was less than 1 ml. Hemolysis analysis done in vitro and in vivo in control (no valve) and experimental (valve) groups used hemoglobin, hematocrit, platelets, plasma free hemoglobin, and lactic dehydrogenase as hemolysis indices. There were no statistical differences. The authors conclude that the CentriSafe valve (Cardiac Systems, Inc., Conshohocken, PA) is safe and prevents fatal backflow and air emboli. The valve is nonthrombotic in anticoagulated blood, can be opened and closed thousands of times, and has a burst strength equal to or greater than other components in the perfusion setup.