Response surface methodology (RSM) is an important tool for process parameter optimization, robust design and other quality improvement efforts. When the relationship between influential input variables and output response is very complex, it’s hard to find the real response surface using RSM. In recent years artificial neural network(ANN) has been used in RSM. But the classical ANN does not work well under the constraints of real applications. An algorithm of regression-based ANN(R-ANN) is proposed in this paper, which is a supplement to the classical ANN methodology. It makes network closer to the response surface, so that training time is reduced and robustness is strengthened. The procedure of improving ANN by regressions is described and the comparisons among R-ANN,RSM and classical ANN are computed graphically in three examples. Our research shows that the R-ANN methodology is a good supplement to the RSM and classical ANN methodology, which can yield lower standard error of prediction under conditions that the scope of experiment is rigidly restricted. Response surface methodology (RSM) is an important tool for process parameter optimization, robust design and other quality improvement efforts. When the relationship between influential input variables and output response is very complex, it’s hard to find the real response surface using RSM. years artificial neural network (ANN) has been used in RSM. But the classical ANN does not work well under the constraints of real applications. An algorithm of regression-based ANN (R-ANN) is proposed in this paper, which is a supplement It makes network closer to the response surface, so that training time is reduced and robustness is strengthened. The procedure of improving ANN by regressions is described and the comparisons among R-ANN, RSM and classical ANN are computed graphically in three examples. Our research shows that the R-ANN methodology is a good supplement to the RSM and classical ANN methodology, which can yield a lower standard error of pr ediction under conditions that the scope of experiment is rigidly restricted.