Recently, Permanent Magnet Synchronous Motors (PMSM) have obtained wide industrial application because that they performance excellently in terms of precision and efficiency. Concurrently, in the field of PMSM control, model predictive control becomes a high performance control method, attributing to its advantages in implementation and multi-objective control. The Finite Control Set Model Predictive Control (FCS-MPC) minimizes a cost function by enumerating optimal voltage vectors (VVs), incurring significant computational burden due to evaluating eight VVs per cycle. A predictive current control algorithm is proposed in this paper on the basis of sliding mode theory to improve traditional MPC. By preselecting VVs, this method substantially reduces the number of VVs evaluated per cycle compared to conventional methods. Simulation results demonstrate the proposed method has comparable dynamic response and steady-state performance to traditional methods with significantly reduced computational requirements.