An optimal design method combining response surface model with genetic algorithm was proposed in the present research to ensure reliable structural parameters of bulkhead. The Box-Behnken design method was selected to determine the location of sampling points for generating the response surface, and the model of maximum stress, total mass property and maximum displacement was established in terms of the input parameters. Then the Pareto optimal solutions were obtained by using multi-objective genetic algorithm to optimize the model. The numerical simulation results show that the optimized total mass of the composite bulkhead is reduced by 23% compared with that of the steel bulkhead, while the maximum displacement is basically consistent with that of the steel bulkhead structure. Therefore, the present research provides useful insight into the structural optimization of composite bulkhead structure in engineering application.