Abstract: Using shaking-flask fermentation in the laboratory to optimize the components of GM1 fermentation medium.On the basis of single experiments and according to the Box-Behnken experimental design principles, the response surface methodology was applied to optimize the three major factors for GM1 liquid-state fermentation.By solving the quadratic regression model equation, the optimal values of the variables were determined as:the molasses 3.24%, complex nitride 0.41%, pH7.91.Then the largest number of bacteria predicted by the theoretical model was 3.0293×108CFU·mL-1.With the optimum medium, the live bacteria number was 3.09×108CFU·mL-1, which increased nearly more than 40% contrasting on the initial culture medium.On the conclusion that response surface methodology was an available way to optimize the fermentation conditions.