In order to solve the problem of large steady-state fluctuation of single-vector model predictive current (MPCC) control for the five-leg dual permanent magnet synchronous motor (PMSM) system, an optimal double-vector model predictive current control strategy is proposed. By using the idea of deadbeat, the time required for two motors to track the reference value of q-axis current is predicted independently. The basic effective vector which can track the q-axis current accurately in a control cycle is screened out. According to the characteristics that the switching state of the common leg must be consistent, the appropriate zero vector is selected to combine the vectors that meet the conditions, and the vector combination that makes the system performance reach the optimum is selected to act on the five-leg inverter to drive the two motors, so as to improve the operation performance of the system. The simulation results verify the feasibility and effectiveness of the proposed double-vector MPCC strategy.