This paper proposed a face recognition method based on principal component analysis (PCA) and genetic algorithm (GA). GA is applied to get the optimal eigenvectors. The ORL and Yale face database was used to test the proposed method, the experiment results shows that the method is superior to the PCA method.