The objective was to optimize the process of extracting effective constituents from Radix Salvia by central composite design/response surface methodol. The effects of influence factors such as solvent fold, ethanol concentration and extract time on extraction rate of tanshinone II a and salvianolic acid were investigated by using box-behnken design and response surface method. The data were imitated using multi-linear equation and second-order polynomial equation. Multiple correlation coefficients from second-order polynomial equation were prior to those from multi-linear equation. The optimum condition of extraction process was 9 fold solvent, 1.5 h for decoction, and 3 times for extraction; and the ethanol concentration was 60%. Bias between observed and predicted values was less than 3%. The optimum model was highly predictive.