Cloud model was used to analyze the spatial-temporal distributions of precipitation based on monthly precipitation data (1960-2012) from 7 stations in source regions of Yangtze River and Yellow River. Results indicate that: the average precipitation in the source region of Yangtze River was smaller than that of Yellow River in the same period, but annual, monthly and non-flood-season precipitation in this region were more concentrated and stable than those in Yellow River. Annual precipitation was more than monthly, flood-season and non-flood-season precipitation, but it was also the most decentralized and unstable; while precipitation in non-flood season was the smallest, it was also the most concentrated and stable. Precipitation as well as its entropy in source regions of Yangtze River and Yellow River during wet and dry seasons gradually decreased from southeast to northwest. Hyper entropy of precipitation was not regular in space, but it showed opposite characteristics in wet and dry seasons in Tuotuohe, Dari and Qumalai. This research could provide scientific reference for the study of climate change in the source regions of Yangtze River and Yellow River.