Abstract: A map layer integration model for automatic mapping of mineral resource potentials,called evidence theory model,was proposed.The model statistically integrated map layers via the evidence theory used in artificial intelligence,and the degree of one locality belonging to the ore-bearing sample set,measured by basic probability function,was used to express the mineral resource potential of the locality.This model could be applied to automatically generating mineral resource potential map in GISs.Drawing a mineral resource potential map with this model needed the following three step procedure.First of all,every map layer was served as the evidence of a knowledge production rule,and whether a locality of the studied area belonged to the ore-bearing sample set or not was served as the corresponding hypothesis of the production rule,and the basic probability function related to the hypothesis was estimated.Secondly,the parallel rule used in the evidence theory was applied to combining all basic probability functions,so a combined basic probability function,which expressed the estimated degree of each locality belonging to the ore-bearing sample set according to the corresponding map layer group,was obtained.Thirdly,as the mineral resource potential map of the studied area,the combined basic probability function map was generated.According to the geological data of the district from Duolanasayi to Ashele in northen part of the Xinjiang Uygur Autonomous Region,the author applied evidence theory model and combined ore-forming belief model to mineral resource target selection.The mineral resource targets delineated by the two models corresponded to each other,which proved the evidence theory model being effective and practical.