Reducing uncertainty is the premise to obtain reasonable analysis of spatial and temporal variation for extreme precipitation events. Based on the methods of quantile mapping and multi-model ensemble, the uncertainty could be diminished and the output data of six general circulation models were downscaled to analyze the spatial and temporal variation of the Extreme Precipitation Indices (EPI) in the Yangtze River Basin. The results indicate that sufficient uncertainty exists in the original global circulation model output, so uncertainty analysis and eliminating its impact by useful measures is necessary. Each EPI tends to increase compared to the baseline period and the maximum increase may reach to about 10%. The EPI shows the consistency of variation only in several regions at temporal scale, while exhibits great difference at spatial scale for the same emission scenario and period