In order to improve the detection performance and the data mining efficiency of data mining-based intrusion detection system,the method of multiple decision tree fusion(MDTF) is to divide a great large dataset into several sub-datasets,mine on sub-datasets by decision tree separately,and detect network data by different sub-decision trees,and then combine the results from multiple sub-de-cision trees by weighted average.Using the dataset of KDD99,the experimental results show that this technique is superior to the single decision tree of mining on a great large dataset for intrusion detection in terms of classification accuracy.Adopting distributed and parallel mining can improve the speed of data mining.