In order to solve the problem that there are many kinds of unknown protocols and the clustering results can not cover all protocols, a classification algorithm based on adjusting mutual information is proposed. By using the improved k-means clustering, the reasonable coverage model of the target sample is constructed, the adjusted mutual information threshold of each clustering center is calculated, the adjusted mutual information value of each clustering center is obtained, and the different clustering mutual information threshold is compared. The effectiveness of the algorithm is verified by experiments. The experimental results show that compared with other traditional classification methods, this method has achieved better results in the classification of binary protocols.