Statistical experimental design strategy (SES) was applied to optimize the medium for the exopolysaccharides (EPS) production of Cordyceps militaris by submerged culture in shaker flask. A significant influence of the glucose and peptone on the EPS production was first evaluated by using a Plackett-Buman design. Then, steepest ascent method was employed to approach the experimental design space. Last, these factors were further optimized using central composite design (CCD) and response surface methodology (RSM). A quadratic model was found to fit the EPS production. The optimum values of the tested variables for the production of EPS were 48.67 g/L glucose, 12.56 g/L peptone, 1 g/L $KH_2PO_4$, 10 g/L yeast extract, and 0.5 g/L $MgSO_4{cdot}7H_2O$. Under optimization of culture conditions, the EPS production was enhanced from 0.78 to 1.96 g/L. In comparison with that of original culture conditions, 2.5 fold increase was obtained.