Computational fluid dynamics (CFD) has become an essential tool for designing and optimizing the structure of blood pumps. However, it is still questionable which turbulence model can better obtain the flow information for axial flow blood pump. In this study, the axial flow blood pump was used as the object, and the influence of the common turbulence models on simulation was compared. Six turbulence models (standard [Formula: see text]–[Formula: see text] model, RNG [Formula: see text]–[Formula: see text] model, standard [Formula: see text]–[Formula: see text] model, SST [Formula: see text]–[Formula: see text] model, Spalart–Allmaras model, SSG Reynolds stress model) were used to simulate the pressure difference and velocity field of the pump. In parallel, we designed a novel drive system of the axial flow blood pump, which allowed the camera to capture the internal flow field. Then we measured the flow field in the impeller region based on particle image velocimetry (PIV). Through the comparison of experiments and simulation results, the average errors of velocity field obtained by the above models are 30.97%, 19.40%, 24.25%, 15.28%, 28.51%, 23.00%, respectively. Since the SST [Formula: see text]–[Formula: see text] model has the smallest error, and the streamline is consistent with the experimental results, it is recommended to use SST [Formula: see text]–[Formula: see text] model for numerical analysis of the axial flow blood pump.