To achieve dynamic performance improvement of the cross-spring compliant micro-displacement magnifying mechanism(CMDMM), response surface methodology(RSM) was used to optimize the structure parameters for multi-objective analysis. Firstly the parametric analysis FEA model was built, and the working principle of the mechanism was analyzed. Having analyzed the kinematics, dynamics with the ANSYS software, the design performance parameters of the mechanism were determined. The standard response surface were established adopting the inscribed central composite designs methods, and the optimize parameters were extracted. The 8 groups of cross-spring thicknesses were selected to be the design variables. The former three steps natural characteristics were taken as optimal objective. The stiffness of the mechanism was used as the constraint function. A multi-objective optimization design was proposed by Nonlinear Programming by Quadratic Lagrangian(NLPQL) method. Three candidate points were given after thousands iterations. One best candidate point was chose to be corrected as the final design point. The whole optimization program was established from all above steps. Compared the former three steps natural frequency of vibration before and after optimization, the first order frequency raised by 80%, the second and third orders become more than 4% increase. From those discussions above, a positive correlation between thethickness of the cross-spring and the dynamic performance can be concluded. It also showed that the thicknesses from different kinematic pairs had different effects on the dynamic. The response surface methodology was effective in dynamics optimization design.