Abstract: In order to identify the characteristics of extreme rainfall under the background of climate change, a joint sample series of annual maximum rainfall and extreme rainfall exceeding threshold was constructed based on the characteristics of rare occurrence, low probability and high rainfall intensity of extreme rainfall events. What’s more, the combined series were fitted by multiple distribution functions such as Generalized Extreme Value (GEV), Gama distribution and normal distribution, and an optimal distribution function was selected to accurately calculate return periods of extreme rainfall. Then the return periods of two heavy rain and flood events in 2003 and 2007 in the Huaihe River Basin were calculated accurately by established method , and the spatial interpolation of the return periods was used to identify the spatial and temporal distribution characteristics of extreme rainfall during the major flood events. The results showed that on July 8, 2007, the rainfall return period in the Huaihe River Basin reached 195-year, with the strongest daily rainfall extremes; in 2003, the maximum rainfall return period was 132-year, and other dates were less than 10-year, so the rainfall coverage was wider. The rainfall characteristics of the two flood events were more distinct, so it is suggested that the corresponding disaster prevention and mitigation measures should be adopted according to different characteristics of floods to minimize disaster losses. The proposed method makes full use of extreme rainfall information and identifies the distribution characteristics of extreme rainfalls, thus it deepens the understanding on severe flood disasters, which can support making strategies of flood prevention and mitigation.