In distribution hydrological models, precipitation is the key input data for analyzing and computing hydrological processes. Finding a way to produce distribution precipitation data is a hotspot in hydrological research. This paper presented the hypothesis that the distribution of precipitation on the earth surface is the result of the effects of atmosphere system and terrain. Moreover, the spatial distribution of natural precipitation is a group of concentric ovals on the flat earth surface, and has a definite centre with maximum precipitation not affected by terrain. Supporting by the hypothesis, this paper established a new precipitation distribution hydrological model which could simulate the spatial distribution of precipitation, and modified the terrain effect on precipitation through Newton interpolation. The position of the precipitation centre and its precipitation amount were simulated in first time, and thus, the model could have a practical value in basin storm analysis and real-time runoff forecasting. The model was tested by the precipitation data of the Xichuan river basin in the Loess Plateau, which indicated that the model had a high precision.