Abstract The study on automated cartographic generalization has little breakthrough due to the lack of depth of generalization knowledge gained nowadays. This paper expounded the methods of intelligent generalization in GIS environment. The key points were: 1)combining knowledge inference with mathematical models; 2) using geographic characteristics oriented mechanism in generalization knowledge, including the knowledge rules orienting to geographic features, to geographic object classes, and to specific geographic regions. In this paper, the authors’ strategy of realizing intelligent generalization in GIS environment was related at first. And then followed the flow chart of math model process driven by knowledge inference, as well as the structure of generalization expert knowledge base. Finally, various geographic characteristic oriented knowledge rules were described.