Abstract: In order to improve the geometric accuracy of the spindle axis of vertical machining center, a structural compensation scheme was proposed, that is, the geometric error of the spindle axis after assembly was obtained by finite element analysis of the whole machine, and the perpendicularity error between the spindle axis and the X and Y axis motion and the parallelism error between the Z axis motion were reduced by adjusting the column predip Angle and the headstock inclusion ratio. Taking a certain type of vertical machining center as an example, sample points were selected by Latin hypercube sampling method, and the geometric accuracy response surface function model of spindle axis was constructed according to the sample point data, and multi-objective genetic algorithm was used for optimization. The results show that with the optimized design scheme, the perpendicularity error between X and Y axes and the parallelism error between Z axis axes are reduced to achieve the purpose of optimization.