Abstract: The quantile estimation values with the various fitted distributeion are generally undetermined because there are three main inference factors: (1) the indetermination of extreme values due to the statistics theory self; (2) the indetermination of the simulation result from globel climate model; and (3) the results from each donwscaling technology. Generally, bigger errors for the estimated quantile come from mixture influence from the above three factors. In this paper, the sampling errors of estimated quantile with the GEV distribution are researched by means of the statistics inference theory. The numerical test for sampling errors of estimated quantile with the GEV distribution in the 156 stations over East China is made. The results show that total number of observations n is the main affection factor and the sampling errors increase with the increasing return period, thus influencing directly the confidence degree of quantile estimation values.