Abstract: Rainfall spatial distribution is the important information in many fields,such as water resource management,drought and flood disaster prediction,and regional sustainable development,but its interpolation has been a puzzle because there are many affecting elements,such as latitude,longitude,elevation,distance to water bodies,slope,etc.,especially in mountainous regions.It is difficult to build a general rainfall interpolation model for different geographical regions.Several kinds of rainfall interpolation methods were introduced in this paper,including global interpolation methods(trend surface and multiple regression),local interpolation methods(Thiessen polygons,inverse distance weighting,kriging and splines),and mixed methods(combined global and local methods).Their advantages,disadvantages and applicability were discussed.Recently,with the development of applied mathematics and artificial neural networks(ANN),some new methods were put forward in the rainfall interpolation,especially the ANN technique,such as Back-Propagation neural networks(BP network) and Radial Basis Function networks(RBFN).Because of the uncertainty of rainfall,more detailed geographical and topographical characteristics are needed to improve the precision of predicted rainfall.Detailed topographical characteristics could be provided by a large scale DEM(digital elevation model) or DTM(digital terrain model),which plays an important role in rainfall interpolation.Different interpolation methods are required in different space or time scales.Even the same rainfall interpolation might get different results in different regions.The mixed methods,combining the advantages of global and local interpolation methods,are useful for improving the interpolation precision,which would be one of the important research fields in the future.