A four-factor three-level Box-Behnken factorial design combined with response surface methodol. (RSM) was used to optimize the medium composition for the recombinant human nerve growth factor (rhNGF) production in Escherichia coli cells, which carried a plasmid encoding NGF DNA as a part of the fermentation strategy for rhNGF. The concentrations of glucose, yeast extract, tryptone and KH2PO4 in medium were changed according to the design consisting of 27 experiments, and the biomass yield was calculated Estimated optimum conditions of the factors for the growth of Escherichia coli cells were as follows: 7.95 g/L glucose, 19.06 g/L tryptone, 13.51 g/L yeast extract and 3.87 g/L KH2PO4. The model prediction of 2.398 g/L biomass at optimum conditions was verified exptl. as (2.357 ±> 0.083) g/L, much higher than any value obtained in initial experiments as well as in studies carried out previously. And the expression yield of the rhNGF product was improved.