In order to improve the yield of L-asparaginase (EC 3.5.1.1, L-ASNase) produced by recombinant Bacillus subtilis, the fermentation medium was optimized by using artificial neural network algorithm. First, the significant factors were screened through single factor experiment and Plackett-Burman experiment. Then the central composite experiment was used to establish the experimental data samples. Finally, a neural network model was established by using JMP10.0 to optimize the composition of fermentation medium. After optimization, the best medium was obtained as follows: sucrose 65 g/L, yeast peptone 28 g/L, corn syrup 11 g/L, KH2PO4 11.5 g/L, NaCl 3.3 g/L, ammonium sulfate 4 g/L, K2HPO4·3H2O 22.5 g/L, MgSO4·7H2O 1 g/L, and L-asparagine 2 g/L. L-ASNase activity reached 515.6 U/mL in the optimal culture medium, which was 90.9% higher than that before optimization. These results presented a basic data for jar fermentation optimization of L-ASNase.