Increasing evidence shows that microbes are important for the protection of human health and the health of other living organisms. At the same time, microbes can cause other organisms to become sick or even die. Through microbe–host interaction, we can understand intuitively the process and mechanism of host infection by microbes. Several methods are developed to predict microbe–host interactions. However, current methods are limited by the cost of interaction verification experiments and accuracy. Therefore, there is still a need for a rapid and accurate method to predict microbe–host interaction. Here, we proposed a novel method based on Integrated Similarity, KATZ measure, and Within and Between Scores (ISKATZWBS) to predict microbe–host interactions. Experimental results show that the proposed method performs well and the AUCs are 0.946, 0.981, 0.954 on the PHI-base, HPIDB, and HMDAD datasets repectively. Compared with other four state-of the-art methods: KATZHMDA, WBSMDA , NGRHMDA and NCPHMDA, the proposed method has higher prediction accuracy.