Abstract: A managerial advisory system,ASSG, for the hilly sandy grassland ecosystems in north China was developed. The system combined empirical knowledge of local experts of ecosystem management and the theoretical research results on soil water dynamics to produce optimal managerial strategies in terms of land use ,planting species/variety and optimal coverage of plants. A three-step interactive inference procedure together with structured knowledge bases was used to guide users through the consulting process step by step. At each step of the inference, users are provided with multiple options derived from a corresponding knowledge base, and the next inferential step is based on users choice in the current step. This interactive scheme of inference gives users maximum flexibility to obtain appropriate consulting results.The main body of the system was implemented in Turbo Prolog, a descriptive language for expert system development. The optimal plant coverage model based on soil water dynamics was encoded into Turbo C and was imbedded into the main body of the program by using the multiple language programming. Sample consultation runs of the system indicated that the system operated smoothly and produced reasonable results of managerial strategies.