Statistical experimental designs were used to optimize the composition of culture media for the production of D-ribose by Bacillus subtilis. A fractional factorial design 2(5-2) was used to determine medium components that significantly affected D-ribose production. The concentrations of glucose and (NH4)2SO4 were the significant factors. Central composite design and response surface methodology were then used to estimate the quadratic response surface and determine the factor levels for maximum production of D-ribose. Finally, the optimal medium composition was obtained (g/L): glucose, 172.75; (NH4) 2SO4, 13.2; yeast powder, 4; corn steep liquor, 8 and MnSO4, 0.5. This optimization strategy increased D-ribose production from 73.21 g/L to 88.57 g/L, an increase of 22% compared with the original conditions. The D-ribose production yield to glucose concentration was also enhanced from 0.37 g/g to 0.52 g/g. Confirmatory experiments were also performed to demonstrate the accuracy of the model. Under the optimal medium using ammonia to control pH in a 5 L fermenter, the D-ribose yield was increased to 95.28 g/L after 3 days of cultivation at 37 ℃.