Study region: The Kaidu River Basin (KRB) is located on the central southern slope of the Tianshan Mountain in Northwest China. Study focus: This work aimed to assess changes and main drivers of snowmelt-driven runoff in KRB associated with three future climate scenarios. Six versions of the “Cemaneige” snowmelt module embedded in the hydrological model “GR4J” were calibrated and evaluated. The bias-corrected climate datasets from CMIP5 Models were used to drive the optimal snowmelt-hydrological model for runoff prediction. The factors that lead to runoff variations were also assessed. New hydrological insights: The significant declining trends of runoff were only predicted in the RCP4.5 and RCP8.5 scenarios. The declining trends of runoff were found in all the seasons. For the annual and summer runoff, compared to the historical period, both the RCP2.6 and RCP 4.5 periods showed a decline in the mid-century and a rise in the end-century; however, RCP8.5 showed a continuous decline during this period. Precipitation and evapotranspiration were ranked as the two most important factors regulating future runoff variations in all RCPs. In contrast, snowmelt timing is the second factor in the historical period, and its importance decreases in the warmer RCP scenarios. These results highlighted that the importance of snowmelt and snowmelt timing to the future runoff depends on the runoff responses to the trajectory of future changes in temperature and precipitation.