The generalized extreme value distribution (GEV) and generalized Pareto distribution (GPD) are used to fit the hourly extreme-precipitation of twelve stations in the Sichuan Basin, based on parameter estimation method of L-moment. Then we compare the effects of the two models. It is appropriate to use Hill plot with D^* to determine the optimal threshold of GPD, and the selected samples are independent. The probability distribution of hourly extreme-precipitation of each station is all in line with GPD and GEV, but GPD model shows better fitting accuracy than GEV model. The maximum hourly precipitation with a given return period was calculated by the two models for each station. The extreme quantiles for 50 a and 100 a return period in Luzhou exceeded 100 mm. In addition to Suining, the relative errors of the estimation of extreme quantiles are basically below 10％ for each model. The estimated results based on GPD are more reliable by the analysis.