[Objective] Studying the optimal probability distribution model of extreme precipitation in order to provide basis for the standard modification of flood calculation method. [Methods] Based on the daily precipitation data from 110 meteorological stations during 1959-2008 in the Huaihe River basin, annual maximum series(AM) and peak over threshold series(POT) were established to compare the applicability of them. The optimal probability distribution models of extreme precipitation for AM and POT were established and the applications were discussed. [Results] In the study of spatial distribution of extreme precipitation, POT was proved to be more reasonable than AM. In dealing with temporal sequence, AM was more reasonable. Checked by K-S method, Wakeby was the optimal function for the two kinds of series. The estimation accuracy of Wakeby was higher than the performance of Pearson Ⅲ, which is regarded as the standard frequency curve in the water conservancy project; especially, the middle part of Wakeby fitted better than its tail did. [Conclusion] The probability distribution model can get a better application in the climate change. The extreme precipitation showed an increasing trend during 1984-2008 and the frequency is increasing. The government needs to take some measures to deal with extreme precipitation disasters.