Computational fluid dynamics (CFD) is beginning to significantly impact the development of biomedical devices, in particular, rotary cardiac assist devices. The University of Pittsburghs McGowan Center for Artificial Organ Development has extensively used CFD as the primary tool to analyze and design a novel axial flow blood pump having a magnetically suspended rotor. The blood-contacting surfaces of the pump were developed using a CFD-based design strategy that involved closely coupling an incompressible Navier-Stokes solver to a parameterized geometry modeler and advanced mesh movement techniques. CFD-based blood damage models for shear-induced hemolysis as well as surrogate functions describing thrombosis potential were employed to help guide design improvements. This CFD-based design approach resulted in the timely development of a pump subjected to five impeller blade refinements, three magnetic gap flow path refinements, and six aft stator blade refinementsall without building expensive physical prototypes and the associated testing for each design iteration. A physical prototype of the final improved pump was fabricated and experimentally analyzed using particle imaging flow visualization. The CFD predicted results correlated well with the experimental data including: pressure-flow (H-Q) performance, detailed flow field features, and particle residence times in the magnetic gap flow path. It is estimated that the present CFD-based design approach shortened the overall design time frame from an order of years to months.