This document presents a proposed hybrid intrusion detection system that combines k-means clustering, k-nearest neighbor classification, and decision table majority rule-based approaches. The system is evaluated on the KDD-99 dataset to detect intrusions and classify them into four categories: R2L, DoS, Probe, and U2R. The goal is to decrease the false alarm rate and increase accuracy and detection rate compared to existing intrusion detection systems. The proposed system applies k-means clustering first, then k-nearest neighbor classification, and finally decision table majority rules. Results show the proposed hybrid approach improves performance metrics compared to existing techniques.