Biometrics is a branch of science that uses unique characteristics or characteristics that exist in individuals of each human being to determine their identity Face recognition has quite interesting challenges, including facial expression recognition. To overcome these problems, use the Convolutional Neural Network (CNN) method. In this study discusses the problem of security systems using biometrics. The flow of system development used in this study is the Waterfall Model where each stage is structurally arranged. Tests were conducted on each of the 100 images on each expression. The test uses epoch value parameters. Conclusions that can be drawn based on the results of testing facial expression recognition systems using Convolutional Neural Network is using the Convolutional Neural Network (CNN) method is able to classify facial expressions quite well and has an accuracy of up to 77% of 100 facial expression images.