Additive manufacturing (AM) is an effective tool for accelerating knowledge gain in development processes, as it enables the production of complex prototypes at low cost and with short lead times. In the development of mechanical circulatory support, the use of cheap polymer-based AM techniques for prototype manufacturing allows more design variations to be tested, promoting a better understanding of the respective system and its optimization parameters. Here, we compare four commonly used AM processes for polymers with respect to manufacturing accuracy, surface roughness, and shape fidelity in an aqueous environment. Impeller replicas of the CentriMag blood pump were manufactured with each process and integrated into original pump housings. The assemblies were tested for hydraulic properties and hemolysis in reference to the commercially available pump. Computational fluid dynamic simulations were carried out to support the transfer of the results to other applications. In hydraulic testing, the deviation in pressure head and motor current of all additively manufactured replicas from the reference pump remained below 2% over the entire operating range of the pump. In contrast, significant deviations of up to 620% were observed in hemolysis testing. Only the replicas produced by stereolithography showed a nonsignificant deviation from the reference pump, which we attribute to the low surface roughness of parts manufactured thereby. The results suggest that there is a flow-dependent threshold of roughness above which a surface strongly contributes to cell lysis by promoting a hydraulically rough boundary flow.