Today’s computer network security systems like IDS, firewall, access control, etc., are not yet 100% trusted, Still they are suffering from the high classification error. Therefore, there is challenge for the researchers to minimize the classification error of the IDS. In this paper, an IDS has been proposed which is based on the decision tree and genetic algorithm. The base of the system is decision tree C4.5 and in the second phase of the intrusion detection system, genetic algorithm is used to overcome the problem of small disjunct in the C4.5. The competence of the system is tested with KDD CUP data set and outcomes of the proposed system are compared with existing systems. It is worth to mention that the experimental assessment of the proposed system is better in comparison to the IDS reported in the literatures.