Decision tree algorithm is a typical classification and regression algorithm, which builds a tree model according to the made training set. Decision tree makes full use of the prior information to automatically select variables and reduce the dimension when dealing with the nonhomogeneous relationship between data, so that the classification results are simple and easy to understand. This paper proposes a network intrusion detection system based on NSL-KDD data set. Improved CART algorithm based on decision tree (DT) is used to classify network attacks. In this paper, the proposed intrusion detection method is tested using independent test data from the benchmark NSL-KDD dataset. The experimental results show that the intrusion detection system based on CART decision tree has good intrusion detection function and efficiency and can meet the application requirements of computer network intrusion detection.