Carbon fiber reinforced polymer/plastic(CFRP) has the properties of anisotropy and inhomogeneity, which can easily lead to processing defects during milling of CFRP. Modeling of milling force is helpful to explore the changing law, optimize processing parameters, and then reduce the appearance of defects. Based on the nonlinear multiple regression method, the paper establishes the functional relationship between the milling force coefficient and the four influencing factors, i.e. the fiber cutting angle, the instantaneous uncut chip thickness, the spindle speed and the axial depth of cut. Results of unidirectional and multidirectional CFRP milling experiments show that the established model can accurately predict the variations of milling forces.