Eigenfaces method, based on principal component analysis, is one of the important algorithms for human face recognition. This paper proposes a retrieval method by comparing the degree of similarity for different candidates using bareheaded photos from college entrance examinations based on eigenfaces. To validate the practicality and efficiency of the proposed method, a human face images retrieval prototype system is developed using eigenfaces and candidate bareheaded photos. The experiment results show that the photos with higher similarity are retrieved more efficiently, which has application value and research significance.