This paper proposes a cascade finite control set model-free predictive control (MFPC) to improve the dynamic performance and robustness against motor parameters mismatch of the PMSM system. For the speed outer-loop, a model-free predictive speed control method that is easy to implement is designed to improve the dynamic performance and reduce the response time. Meanwhile, the difference in the sampling period for the outer-loop and inner-loop is considered and compensated. Furthermore, a finite control set MFPC strategy is designed for the current inner-loop, which can eliminate the dependence on motor parameters for the predictive current control. Therefore, the robustness against parameters mismatch will be improved. Finally, the simulation is carried out to evaluate the performance of the proposed cascade MFPC by comparing it with the conventional PI and MFPC.