The intrusion detection system (IDS) is one way of protecting a computer network. This kind of technology enables users of a network to be aware of the incoming threats from the Internet by observing and analyzing network traffic. The proposed technique involved four steps, first apply DBSCAN clustering which is used to make clusters, based on this obtained clusters we trained the network with by Back Propagation algorithm. We also apply Information Gain based Feature Selection method to identify the important features of the network. We trained the network once with all features and then reduced features this shows that we attain high detection rate and in efficient time. The developed network is used to identify the occurrence of various types of intrusions in the system. The performance of the proposed approach is tested using KDD Cup’99 data set available in the MIT Lincoln Labs. Simulation result shows that the proposed approach detects the intrusions with high detection rate and low false alarm and in high efficiency in terms of time.