Based on climate change scenarios derived from 8 GCMs (Global Climate Models) and daily precipitation data during the period of 1961-2015 in Shanghai,a cumulative distribution function-transform (CDF-T) model was developed to downscale the daily precipitation on the meteorological observation site scale.The results showed that this downscaling method can improve the simulation results,which has more rain days,lower precipitation intensity and less precipitation.It shows that using the daily data in flood season to develop downscaling model can improve the CDF curve,the total amount and intensity of precipitation in flood season compared with that using whole-year daily data.Similarly,this method can improve the correlation of the observed and correct mean value of the days,amount and intensity of the rainstorm as well as the daily maximum precipitation in longer return periods.For the period of 2016-2095,it was found that the precipitation and its intensity will increase,while the rainy days both for the whole year and flood seasons will decrease in Shanghai,compared with the current stage (2006-2015).There is likely to have more drought and flood events and intensify extreme rainfall events with the increased average and extreme values of rainstorm.The daily maximum precipitation of the recurrence intervals over 50 years will decrease in the former 40 years and increase in the later 40 years in the future.Consequently,the downscaling model of CDF-T can be applied in meteorological observation site scale and provide the downscaling method and climate data for climate change projection and assessment.