In this paper, a two-stage approach is proposed for optimizing the robust injection molding parameters. In the first-stage optimization, the simplex method is used to search for qualified parameters when the starting parameter is disqualified. In the second-stage optimization, forgetting factor recursive least square algorithm is used to build the model between process parameters and quality index. Robust parameters are obtained by using the steepest ascent method based on the model. Time and experiment costs are saved by using the two-stage approach. Experiments and results on mold flow analysis software show its validity.