Climate change is one of the major global issues commonly concerned by international communities. Taking the Three Gorges Reservoir which is the largest water conservancy project in China as the study area, this study attempts to evaluate the variability of hydro-climatic elements including precipitation, temperature, relative humidity and sunshine hours during past 49 years (1961-2009) based on observed meteorological data, and predict the evolution trends of precipitation, temperature, evapotranspiration and runoff in the future 30 years (2021-2050) under climate change based on a distributed hydrological model and climate model outputs. Specifically, the temporal variations and sudden changes of precipitation, temperature, relative humidity and sunshine hours during past 49 years in the Three Gorges Reservoir are analyzed combining moving-average and linear regression with Mann-Kendall method, and the precipitation and temperature data in the future 30 years are obtained from the average dataset of 20 global climate models using a statistical downscaling model, and the evapotranspiration and runoff data in the future are obtained based on the validated hydrological model. The results indicate that: 1) during 1961-2009, the precipitation slightly decreased and the estimated sudden change time was 1973, the temperature significantly increased and the estimated sudden change time was 1963, the relative humidity decreased but there was no sudden change time estimated, the sunshine hours significantly decreased and the estimated sudden change time was 1967; 2) in the future, comparing with the historical average, the temperature in the region will increase by 1.3°C and the evapotranspiration will increase by 2.8%, while the precipitation and runoff will decrease by 0.8% and 8.2% respectively. Although the precipitation will not change greatly, the reduction extent of runoff is larger than that of precipitation, bringing forward higher requirements for integrated water resources management in the Three Gorges Reservoir.