Based on the observational precipitation data from Hangzhou Climate Observatory and the simulated precipitation data of the global climate model CMIP5, the long-term trends of precipitation variation in Hangzhou City are investigated. Moreover, the risks of extreme precipitation in Hangzhou City under future climate scenarios are estimated by using a downscaling method that adopts the cumulative distribution function transform (CDF-T). The results show that there is no significantly increasing or decreasing trend in centennial precipitation series (1907-2015) at Hangzhou Climate Observatory, but the precipitation in spring obviously decreases with a rate of 32.1 mm·（10 a）-1 and the winter precipitation significantly increases with a rate of 35.4 mm·（10 a）-1 during 1980-2015. The extreme precipitation in Hangzhou City has been intensifying, which can be interpreted by that the maximum values of 3 h, 6 h and daily precipitation have been increased and the return periods of extreme precipitation have been shortened. Du-ring the period from 1988 to 2015, the values of the 3 h, 6 h and daily precipitation of each return periods are higher than that of the period from 1961 to 1987. An extreme daily precipitation whose probability is once in 100 years during 1961-1987 has become an event with probability once in 50 years or once in 20 years during 1988-2015. The CDF-T downscaling analysis of simulated precipitation of the CMIP5 models indicates that the occurrence probability of extreme precipitation in Hangzhou City during 2020-2039 might increase under all of the RCPs future climate scenarios. In the period from 2020 to 2039, the occurrence probability of the extreme precipitation over R95p and R99p in Hangzhou City will be 11.08 d·a-1 and 2.24 d·a-1 respectively. Compared with the average values of current climate, the occurrence probability of the over R95p and R99p is increased by 3.52 d·a-1 and 0.69 d·a-1, respectively.