In modern era, due to advancement in technology, internet users are growing rapidly. With this pace of increase in internet users, security has becoming a major challenge. It is essential to deal with such security vulnerabilities for which, intrusion detection system has become a necessity. Intrusion detection system is categorised as network based and host based. Network based intrusion detection system collects all the packets which travel over the network and groups them in order to find patterns of attack. Neural networks are prominent for being adaptive, which dynamically modify themselves as they learn from initial training. On careful analysis of the data, many such attacks can be identified. Using neural networks, this malicious activity over the network can be effectively identified. With an exhaustive analysis carried out, an intelligent system to find such patterns in order to maximize the recognition rate using various neural network methods such as classification, pre-processing and pattern matching to improve the performance is proposed. © 2016, International Journal of Pharmacy and Technology. All rights reserved.