Prevention and mitigation of landslide disaster have received increasing attention due to frequent landslides and their severe threats to the lives and properties of human being. Risk assessment of landslide is a critical work. Fuzzy comprehensive evaluation is a common method of researching the risks of landslide, and assigning weight to each index is a difficulty. In this article, fuzzy comprehensive evaluation based on information entropy, which has been successfully applied in other fields, is introduced into the risk assessment of landslide. Information entropy is used to assign weight to each index and the result of assigning weight is more objective with no subjective randomness. Moreover, vulnerability indexes are selected in addition to stability indexes, which avoid the evaluation result of “disasters of no harm” and finally make the evaluation result more logical. With Shennongjia area as a case study, the model of fuzzy comprehensive evaluation is applied with weights assigned by the theory of information entropy to evaluating the risks of five typical landslides. The evaluation result is consistent with the actual situation, hence can be used in risk assessment to provide basis for the prevention and mitigation of landslide.