To solve the problem that the uncertainty of design variables is usually ignored in the process of optimization design in milling， which may lead to the violation of optimization design constraints， the multi-objective robust optimization design of milling parameters is studied based on the approximate model method. In the proposed method， the maximum milling force and the surface roughness after milling are the objective functions， and the milling linear speed and feeds per tooth are the design variables. The polynomial response surface model and radial basis function model are used to replace the implicit relationship between the objective functions and design variables. NSGA-Ⅱ multi-objective genetic algorithm combined with the Monte Carlo simulation method is used to compare and analyze the conventional deterministic optimization for milling parameters and the robust optimization considering the fluctuation of design variables. The results show that the method can effectively reduce the milling force in the milling process and improve the surface quality of the workpiece after machining with high reliability.