Terminal-sensitive projectile is an advanced new-type of ammunition, having complex struc?ture and a multiplicity of influencing factors, the overall optimal design to the terminal-sensitive projectile system is thus a very difficult work. Neural network and genetic algorithm were used in this paper, because of the highly non-linear reflecting capability of neural network and the overall optimizing ability of genetic algorithm. Working principle of terminal-sensitive projectile was introduced, and a project of optimal design was confirmed. Using theory of neural network and orthogonal experiments, a simulating model of termi?nal-sensitive projectile system efficiency was established. On this basis, using a hybrid genetic algorithm, an optimal design to the neural network simulation model was carried out. From it an optimal arrangement of several main factors affecting the system efficiency was obtained. Validation to the above optimal results was then done, and some conclusions about this optimal design formed, showing that the optimal arrange?ment of those influencing factors are wholly in accordance with the actual state, and that the method can provide scientific foundations for the future efficiency research of terminal-sensitive projectile systems.