Abstract: In order to obtain the optimal performance of the switched reluctance motor (SRM) drive system, this paper proposes the system-level deterministic and robust optimization design methods, and optimizes the parameters of the motor and controller simultaneously. For the system-level deterministic optimization method, the multilevel optimization technique, the orthogonal design of experiments and the approximate model are employed to reduce the calculation cost, and a genetic algorithm is used to search for the optimal solution. Furthermore, considering that the motor is inevitably affected by uncertain factors such as manufacturing error in mass production, a sequential Taguchi method is applied for the robust optimization of the motor system. By comparing and analyzing the performance of the optimal solutions obtained by the two system-level optimization methods, it can be found that the system-level deterministic optimization can obtain the optimal electromagnetic performance, but its robustness is poor. The reliability of the system-level robustness optimization is higher, which can effectively reduce the failure rate of the motor in the batch production process while improving the overall performance of the motor drive system. Finally, a six-phase 12/10 SRM is manufactured. The experimental results verify the accuracy and feasibility of the robust optimization method.