Abstract: Products shrinkage is an important index in measuring part quality. A method of combining neural network and genetic algorithm was proposed to optimize the volumetric shrinkage in the short fiber-reinforced composite injection molding process. In this study, the objective function is a minimum problem of the volumetric shrinkage. The design parameters include fiber aspect ratio, fiber content, mold temperature, melt temperature, holding time and holding pressure. On the basis of orthogonal experiment design, Moldflow software is applied in the short fiber-reinforced composites injection molding process. A neural network model is developed on the basis of the simulation results to map the complex nonlinear relationship between design parameters and the volumetric shrinkage. The GA is interfaced with this predictive model to improve the volumetric shrinkage significantly by optimizing the design parameters. The optimized parameter combination is obtained. Finally, the part volumetric shrinkage is eliminated by using the optimized parameters.