In vitro testing of hemolysis is essential for the validation and development of ventricular assist devices. However, as many factors influence hemolysis, such tests’ inter- and intralaboratory reproducibility is poor. In this work, CentriMag blood pumps were used to conduct a hemolysis study according to ASTM F1841 with blood from 23 bovine donors. Complementary blood analysis, including cell count, plasma composition, and viscosity, was performed to identify factors relevant to the variability of hemolysis testing results. Three strategies were tested to improve reproducibility: albumin supplementation, maintaining glucose concentration, and replacement of plasma with plasma-like buffer. Differences in red blood cell stability among donors were responsible for the largest portion of the total variance. Hematocrit varied widely among donors, and its adjustment to a standard value led to the artificial introduction of between-donor differences, especially in viscosity. It seems likely, that a more careful selection of donors with similar characteristics or repeated blood collection from the same donor could improve reproducibility. However, no direct correlations were found between the susceptibility to hemolysis and individual donor or blood characteristics in this study. The addition of albumin and glucose had a negligible effect while washing blood samples with artificial plasma significantly reduced mean hemolysis, although not its variation. The findings contribute to the understanding of variability in hemolysis experiments and give reason to question the common standard operating procedures, such as hemodilution or maintaining glucose concentration. To confirm the factors identified here, additional studies isolating the effects of individual factors are necessary.