Abstract: This research presents an intrusion detection method for network datasets using Minimax Probability Machines (MPM) and genetic algorithm. The minimax probability machines can achieve the comparative performance with the Support Vector Machine (SVM). To do more accurate data classification and decrease the training time of classifier, we present a genetic feature optimized method for minimax probability machines. Genetic algorithm is used to optimize the feature so as to generate newly features to boost minimax probability machines do more accurate classification and need less training time. A new classifier model based on minimax probability machines with genetic feature optimized is proposed and is applied to intrusion detection in this paper. The experimental results show that the classification method with genetic feature optimized has better performance than the traditional learning method.