Abstract: For improving classific precision of network intrusion detection model，reducing the number of training data set and learning time，Based on Support Vector Regression（SVR），a new supervisal algorithm using Lagrange Support Vector Regression（LSVR） is proposed.The experimental results using KDD CUP 1999 data set show that LSVR has better generalization ability，quicker iterative speed，higher detection accuracy，and lower error rate than SVR.