Effective fault diagnosis methods can improve the reliability of the two-level inverter used in traction systems. In this paper, a diagnosis method for open-circuit (OC) faults in transistors is proposed, which is based on finite control set-model predictive control (FCS-MPC). First, the cost function is constructed and the inverter is controlled by the FCS-MPC. Then, the actual and estimated value of the cost function are calculated. the difference between them are applied to fault detection. Next, Combine the command signals and current direction, the numbers of affected sampling points in a period are counted, different numbers correspond to different fault locations. The numbers are normalized and the largest percentage is used to determine fault location. Finally, the simulation platform is established, the feasibility and effectiveness of the proposed fault diagnosis method is verified by simulation results.