Cyber security attacks have become increasingly sophisticated and complex. Intrusion detection is one of the major challenge in computer security that aims to detect unseen attacks in order to protect internal networks. Intrusion detection system compact with huge quantity of data which is a crucial task to keep good quality of features and remove the redundant and irrelevant features. An intelligent intrusion detection system combining ensemble classifier for classifying abnormal and normal actions in the computer network is presented. In this work, K- Nearest neighbourhood algorithm is used to determine the best optimal subset. The misuse detection model is built based on the C5.0 algorithm. Further, anomaly detection model is implemented by one-class Support Vector Machine (SVM) to detect the anomalies. Integration of ensemble algorithms helps achieve better classification. Based on the exhaustive analysis carried out, we propose that on incorporating, NSL-KDD Dataset, which is a well known benchmark for intrusion detection dataset, the system will result better in terms of anomaly detection rate, false positive, false negative, true positive, true negative, and f-score. © 2016, International Journal of Pharmacy and Technology. All rights reserved.