Abstract: Against the defect of the traditional cartographic generalization strategy which processes the map features separately hence often causes spatial and semantic conflicts between different geographical features on the map, we take into account the spatial relevance between terrain and rivers to introduce a cooperative generalization method which integrates the contour clusters with the river networks. Using the spatial proximity relations between the vertexes of the above-mentioned two map features in the constrained Delaunay TIN and via extracting and organizing the bends of valley contours, we construct the cooperative relation between the two features. On this basis, we divide contour clusters into different segment types according to the local topography, and then perform the corresponding simplification under the subset of the river networks given by a hierarchical selection method, so as to keep the position of every river of the networks to the bottom of the valleys. The experiment result proves that this cooperative method can effectively avoid the conflicts between the contour clusters and the river networks, and helps to improve the intelligence level of cartographic generalization.