Abstract In this work, a 22 factorial design was employed combining with response surface methodology (RSM) to optimize the medium compositions for the production of alkaline β-mannanase by alkaliphilic Bacillus sp. N16-5 isolated previously from sediment of Wudunur Soda Lake in Inner Mongolia, China. The central composite design (CCD) used for the analysis of treatment combinations showed that a second-order polynomial regression model was in good agreement with experimental results, with R2=0.9829 (P<0.05). The maximum activity was obtained at NaCl concentration (84.4g l−1) and sodium glutamate (3.11g l−1) and a high medium pH around 10.0. Under such conditions, the activity of alkaline β-mannanase achieved 310.1U/ml in the scale of 5-l fermenter, which was increased nearly twice compared with the original. Through optimization, the substrates shifted from the expensive substrates, such as locust bean gum and peptone, to the inexpensive ones such as konjac powder, soymeal, and sodium glutamate. The experiment results also suggested that the environmental conditions of high salinity and high alkalinity, as well as the inducer substrates, play very important roles in the production of the alkaline β-mannanase by alkaliphilic Bacillus sp. N16-5.