Abstract: Under the guidance of Integrated Orderly Classification System of Grassland (IOCSG), interpolation methods of Analytic Method based on Multiple Regression and Residues (AMMRR), in conjunction with DEM and meteorological data as main data source including annual accumulative temperature (), annual precipitation (r) of 113 stations in Gansu Province, were applied in spatial interpolation. Then the grassland classes in Gansu were classified through spatial overlay analysis. The results showed that the grasslands could be classified into 28 classes in whole province. Four classes with the largest area, i.e. ⅢA3 (Cool temperateextrarid temperate zonal desert), ⅢB10 (Cool temperatearid temperate zonal semidesert), ⅢE31 (Cool temperatehumid forest steppe，deciduous broad leaved forest) and ⅠF36 (Frigid perhumid rain tundra，alpine meadow), were distributed in north mountains, Hexi Corridor, Loess Plateau and Gannan Plateau. Meanwhile, four classes located in subtropics had the smallest area accounting for less 1%. The results also showed that the distribution of grassland classes fitted to the climatic characteristics of regions in Gansu. The interpolation of and r was closer to reality and the map of grassland classes could fully reflect the zonal patterns of grassland vegetation distribution, which proves that the AMMRR method has a better simulative result because the effect of topographic factors (longitude, latitude and elevation) to climatic factors was covered.