Phenomenological studies on mechanical hemolysis in rotary blood pumps have provided empirical relationships that predict hemoglobin release as an exponential function of shear rate and time. However, these relations are not universally valid in all flow circumstances, particularly in small gap clearances. The experiments in this study were conducted at multiple operating points based on flow rate, impeller speed, and tip gap clearance. Fresh bovine red blood cells were resuspended in phosphate-buffered saline at about 30% hematocrit, and circulated for 30 min in a centrifugal blood pump with a variable tip gap, designed specifically for these studies. Blood damage indices were found to increase with increased impeller speed or decreased flow rate. The hemolysis index for 50-microm tip gap was found to be less than 200-microm gap, despite increased shear rate. This is explained by a cell screening effect that prevents cells from entering the smaller gap. It is suggested that these parameters should be reflected in the hemolysis model not only for the design, but for the practical use of rotary blood pumps, and that further investigation is needed to explore other possible factors contributing to hemolysis.