Abstract: In order to improve the detection accuracy of the tire dynamic balancing machine, the response surface method and multi-objective genetic algorithm are used to optimize the main shaft system of a tire dynamic balancing machine. Firstly, the static analysis and modal analysis of the spindle system model are carried out. The size parameters of the spindle are taken as the design variables, and the optimization objectives are to reduce the deformation of the spindle system, increase the natural frequency and reduce the mass. Secondly, the design variables with the greatest correlation with the optimization goals are identified by parameter correlation analysis. Finally, the response surface model is established and the multi-objective parameter optimization is carried out. The results after optimization show that the maximum deformation is decreased by 3.9%, the first-order natural frequency is increased by 16.2%, the spindle mass is decreased by 23.0%, and the spindle system mass is reduced by 5.5%. The optimization effect is remarkable, the dynamic and static characteristics of the spindle system is optimized, and through experimental verification, the spindle optimization improves the detection accuracy of the tire dynamic balancing machine.