For the defects of under filling, heavy forming load and high peak value of local stress in the forging process of automobile generator pole, the generator pole forging process was numerically simulated by the software of Deform. For the optimal objective to reduce unfilled rate, forming load and peak value of local stress, and the optimal variables with blank diameter, forging temperature, extruding speed and friction condition, the second order response surface model (RSM) of three optimal objectives was set up by central composite experiment design (CCD) with four factors and five levels, and the analysis of variance (ANOVA) of RSM was carried out. Furthermore, the variance analysis was performed on the established model, and its reliability and accuracy were verified by simulated tests. The practice production indicates that generator poles with less forming defects can be obtained by combining the optimized process parameters, and a high forging accuracy of 0.5 mm can be achieved. Therefore, the peak values of forging load and local stress are deduced,the die life is prolonged and the production efficiency is improved.