Abstract: Geographic information system (GIS) is a new technique successfully applied to g eological research. It is advantageous in processing data with spatial distrib uti on. The evidence weight model is a mathematical model combined with GIS techniqu e. Based on the spatial relation of elemental data, it synth esizes various favorable ore-forming factors (evidence layers) and can be used in mineralization prognosis and evaluation efficiently. The northwest of Sichua n Province is an important micro-fine disseminated gold resource base in China. In t his sense, it is significant to have a metallogenic analysis of the region and t o search for the areas favorable for gold mineralization. This paper is an empi rical example for application of evidence weight model in the northwestern Sichu an Province. 　　Based on the comprehensive analysis of features of gold mineralization in this r egion, five databases have been created: geology database, geophysics database, geochemistry database, remote-sensing database, and ore d eposits (occurrences) database. According to the analysis of the ore-contro lling factors in this region, we selected 10 ore-producing e vidence layers, which include the stratigraphic lithological assemblage, lin ear faults, minor plutons, joints of structure from remote sensing, geochemical anomalies of Au, As, Sb, Ag, and the gravity and magnetic anomalies d educed to different elevations. Extracting the data from the above mentioned fiv e databases, the authors calculated the evidence weight values and the correlation degrees be tween the particular evidence layers and ore-deposits (occurrences), created a n e vidence weight model of gold mineralization in the northwestern Sichuan Province . Us ing the evidence weight model in this paper, the authors analyzed the features o f gold mineralization in this region, indicated existence of four main gold min eralization zones in four different directions: the NW-directed gold mineralization zone controlled by Maqu-Lueyang structure belt, the NE-directed gold mineralizatio n zone controlled by Q ingchuan-Pingwu structure belt, the SN-directed gold mineralization zone cont rolled by Minjiang structure belt, and the EW-directed gold mineralization zone cont rolled by Xueshan structure belt. Based on the analysis of distribution charac teristics of gold mineralization zones and spatial coupling relationship s among them, the favourable areas have been predicted in three levels accordi ng to the ore formation probability values. Finally, some suggestions related to application of GIS in metallogenic prognosis are also proposed.