This paper presents a methodology that combines sensitivity analysis and multi-objective optimization in order to design the scheme of a bearingless permanent magnet synchronous motor (BPMSM). This approach is adopted in order to address the problem that, in the optimization of the design of BPMSM, it is not possible to consider accuracy and efficiency simultaneously. Initially, the preliminary scheme of BPMSM is designed using the magnetic circuit method. Subsequently, the response surface method is utilized to construct the BPMSM approximation model. Following this, the sensitivity analysis of the motor model is employed to select the primary optimization variables, thereby enhancing the optimization efficiency. Finally, the multi-objective optimization based on finite elements is conducted to ensure the calculation accuracy. The finite element results demonstrate that the optimized motor’s volume is reduced by 12.63%, and the radial suspension force is augmented by 26.80% under the condition of meeting the performance targets. This substantiates the viability of the multi-objective optimization design method employed in the BPMSM.