The effect of hematocrit (Ht) on in vitro hemolysis test results (i.e., index of hemolysis) was evaluated using a Baylor/NASA prototype axial flow pump. Red blood cell suspensions of six different Ht (5, 10, 15, 20, 30, 40%; n = 30) were prepared and used for this evaluation. The pump was operated for 60 min under 5 L/min flow conditions, and blood samples were taken every 10 min to measure plasma free hemoglobin levels. The normalized index of hemolysis (NIH) was calculated using the regression line slope between time and plasma free hemoglobin level, and relationships between NIH and Ht or hemoglobin (Hb) were checked. NIH and Ht had a statistically significant (p < 0.0001) correlation with a coefficient of fit of 0.976, and NIH and Hb had a statistically significant (p < 0.0001) correlation with a coefficient of fit of 0.976. To reduce the effect of Ht, NIH/Ht was proposed and compared with a modified index of hemolysis (MIH), which was normalized by the Hb level of blood. NIH/Ht and MIH had a poor correlation with Ht (coefficient of fit, 0.608) and Hb (coefficient of fit, 0.577), respectively. When blood that has a wide range of Ht or Hb values is used for in vitro hemolysis tests, NIH/Ht is suggested for use as an index of hemolysis to evaluate the hemolysis characteristics of rotary blood pumps because MIH has no dimension and it requires Hb values. In contrast, NIH/Ht has a dimension of g/100 L, which is quite understandable, and it does not require measurement of Hb levels of blood; it is therefore cost-effective.