Pump manufacturers recommend setting roller pump occlusion such that the level of a 100 cm column of crystalloid drops 2.5 cm/min (Sarns, 8000 Modular Perfusion System, operator's manual, roller pump software version 2.3L. May 1993; 2.1-2.14). Though this almost occlusive setting ensures accurate pump flow, it has been shown to cause more hemolysis than nonocclusive pumps (Noon GP, Kane LE, Feldman L et al. Reduction of blood trauma in roller pumps for long-term perfusion. World J Surg 1985; 9: 65-71). We conducted a clinical study (n = 19) to compare the standard occlusion method with the dynamic method and to determine the accuracy of flow for the nonocclusive pump. Standard occlusion was set by clamping the pump tubing distal to the arterial line filter and timing the drop in pump outlet pressure as indicated by a pressure transducer connected to the filter. The occlusion setting, expressed in mmHg/s, was recorded for each roller at two specific points along the raceway. The pump was then set nonocclusively with the dynamic method using the Better Header (BH) (Circulatory Technology, Oyster Bay, NY, USA). Readings of the change in pressure in the same two selected points on the raceway were taken. The latter was repeated after discontinuation of bypass. Flow was recorded throughout the procedure from both roller pump output display and a flow meter (Model #109 Transonic, Ithaca, NY, USA). The average drop in pump outlet pressure for the standard method was 1.3 +/- 4.0 (range 0-18 mmHg/s), and for the dynamic method was 38 +/- 28 (range 1.2-89 mmHg/s). Off bypass, the average reading was 44 +/- 38 (range 2.0-103 mmHg/s). Regression analysis indicates that patient flow, when corrected for retrograde flow by the dynamic method, equals 1.003 x revolutions per minute + 40 ml/min (r2 = 0.964). The average error between indicated pump flow, corrected for retrograde flow, was -1% (range from -6.7 to 6.6%). We conclude that the BH allows nonocclusive setting (30 times less than our standard method) without sacrificing pump flow accuracy.