Understanding the climate change impacts on water and carbon cycles is of great importance for comprehensive watershed management. Although many studies have been conducted on the future climate change impacts on either water cycle or carbon cycle, the potential impacts on water-carbon coupling cycles are still poorly understood. This study used an integrated hydro-biochemical model (SWAT-DayCent) to quantitatively investigate the climate change impacts on water-carbon coupling cycles with a case study of typical loess hilly-gully watershed-the Jinghe River Basin (JRB) on the Loess Plateau. We used climate scenarios data derived under the three Representative Concentration Pathways (RCPs2.6, 4.5 and 8.5) by five downscaled Global Circulation Models (GCMs) and set two future periods of 2020–2049 (near future, NF) and 2070–2099 (far future, FF). It was projected that the annual precipitation would generally decrease slightly during the NF period but increase by 4–11% during the FF period, while the maximum/minimum air temperatures would increase significantly. The average annual streamflow would decrease (with up to 20.1% under RCP8.5) and evapotranspiration (ET) would remain almost unchanged during the NF period; however, both of them would increase during the FF period. The net primary production (NPP) would be generally higher due to the CO2 fertilization, whereas the soil organic carbon would decrease across all scenarios due to the warmer climate. The NPP-ET was projected to be closely coupled across all scenarios, and this coupling was mainly controlled by the inter-annual variability (IAV) of precipitation. Moreover, the precipitation IAV combined with NPP-ET coupling could also jointly control the NPP variability in the JRB. These projections in water-carbon coupling cycles can be useful to make better-informed decisions for future water resources and ecosystem management of the loess hilly-gully regions.