Against the defect that the traditional computer network security management system processes a mass of data with low efficiency and accuracy, a novel network security management model based on data mining is proposed in this paper. It makes use of multi-source data collection strategy to acquisition and integrate the relevant data of different security products, and takes advantage of data mining technology for massive data to analyze them comprehensively and intelligently and response automatically. The experimental result reveals that the model performs well. It can identity the real attacks from a large number of security events with a good performance, so that the alarm information will be refined with low false or wrong alarms, the accuracy, intelligence and adaptability of network security management will be enhanced and the requirements of new security situation will be met.