In this paper, we propose Genetic Algorithm (GA) to improve Support Vector Machines (SVM) based Intrusion Detection System (IDS). SVM is relatively a novel classification technique and has been shown higher performance than traditional learning methods in many applications. So several security researchers have proposed SVM based IDS. We use fusions of GA and SVM to enhance the overall performance of SVM based IDS. Through fusions of GA and SVM, the "optimal detection model" for SVM classifier can be determined. As the result of this fusion, SVM based IDS not only select "optimal parameters" for SVM but also "optimal feature set" among the whole feature set. We demonstrate the feasibility of our method by performing several experiments on KDD 1999 intrusion detection system competition dataset.