Crashworthiness is one of the most important characteristics in automobile design, which is inflected by a large number of different factors, the influences of all these factors on crashworthiness are not identical. In order to screen the factors which influence the crashworthiness of the automobile most remarkablely, the graphics analysis method of the unreplicated saturated factorial design was employed. It is more and more widely used in the screening of factors because of its huge benefit in economic and the technical reason. The paper applied half-normal probability plot method to get the active and sensitive factors, and D-optimal design was employed to choose design sample points. The full quadratic polynomial approximation model was used to construct the response surface model of the internal energy. Finally, Multi-Island Genetic Algorithm was utilized to obtain the optimal solution of the multivariable crashworthiness optimization of the vehicle body under the full-scale frontal impact load.