Aiming at the lack of the traditional design method of the forging die, the response surface methodology based on design of experiments is proposed to optimize the parameters of flash gutter in forging die. According to the central composite inscribed experiment design, numerical simulation was carried out. On this basis, the quadratic polynomial response surface model was built by regressive analysis. The optimized flash gutter sizes are obtained by analyzing the response surface model with an objective of minimizing flash volume. The forging die of automobile flange has been studied as an illustration to validate the application of the response surface methodology. Finally, the comparison was done using genetic algorithms and the results show that the response surface methodology is better than the genetic algorithms with shorten time and better effect when the factor is small.