In this study, the 2DPCA feature extraction method and SOM classification were used to classify human sex based on facial images. The trial was conducted on 100 facial images consisting of 50 male facial images and 50 female facial images. The 100 facial image data used were divided into 70 training data and 30 test data. Based on the test results, the best accuracy obtained is 86.67% using a start learning rate of 0.6 and a maximum iteration of 10000. level of accuracy, where a larger maximum iteration value results in a higher level of accuracy.