Abstract: According to the theory of expert system and the principle of complex classification of grass land, a prototype of grass resource investigation expert system by remote sensing is developped. The system——GRIES (Grass Resource Investigation Expert System) has the same structure as a usual expert system. The authors design an model of uncertain inference to solve the uncertainty and uncomplety of the knowledge and the data. With the experiments of some examples, a combination of the technique of expert system and remote sensing can make full use of various auxiliary-information to overcome the disadvantages existing in maximum-likelyhood classification. At the meanwhile, GRIES can take the advantage of the knowledge and experiences of the experts to improve the accuracy of classification. After some changes of the knowledge-base and the data-base, GRIES can also be used in the areas such as vegetation field analysis, landuse investigation, predication of growth of crops and other kinds of resources investigation by remote sensing.