Joint modeling of exponentially fitting and least square method was proposed for key technical problems on milling temperature and heat deformation of tool in virtual numerical control (NC) simulation. Mathematic model of surface temperature and its influencing factors based on ball-end mill was established. Correlation coefficients of model are over 0.8, and the errors comparing with the experimental ones are within 10%, which shows that the model meets precision requirement, and fitting effect is good. By this model a mathematical model of the ball-milling cutter’s thermal deformation was established, and the relative error is small by comparing with test values, which proves that this model has high calculation precision and practical meaning. A simulation prediction system of milling temperature and ball-end mill’s heat deformation in virtual NC was developed based on above mentioned models, which can provide theoretical basis for machining error prediction, optimizing milling parameters and improving machining quality.