In this paper，repeated incremental pruning to produce error reduction which is a rule learning ethod is used to solve network traffic classification. The method can be used to dig out the correlations between ttributes and classes，which are utilized to build a classifier for traffic classification. The proposed method can ecrease the classification error rate when the traditional machine learning method has a large number of imbalanced ata sets in the network traffic. Experiments show that the method has a very high classification of accuracy，recall and recision in network traffic classification standard data sets.