This study investigates the projected changes in runoff (R), precipitation (P), evapotranspiration (ET), and soil moisture (SM) under CMIP5 and CMIP6 scenarios, quantifying uncertainties at annual and seasonal scales. The results show that all four variables increase over most of the global land under a high emissions scenario (SSP5-8.5) during 2080–99, with 72%, 81%, 82%, and 66% of global land area showing increased R, P, ET, and SM, respectively. Model uncertainty dominates the total projected uncertainties, with 76% (R), 73% (P), 89% (ET), and 95% (SM) of uncertainties in the 2090s. Internal variability decreases over time, while scenario uncertainty increases. Low-latitude regions have the greatest uncertainty in hydrological projections. The uncertainty in P contributes the most to R uncertainty, with 93% at annual scale. CMIP5 and CMIP6 models show similar performance in terms of hydrological changes and uncertainty composition. The study provides a theoretical basis for improving hydrological components in global climate models. The results highlight the importance of quantifying uncertainty sources in hydrological projections to improve the reliability of climate models and future water condition projections. The study also compares CMIP5 and CMIP6 results, finding that CMIP6 has higher model agreement in some regions but larger uncertainties in long-term projections. The contribution of scenario uncertainty is higher in DJF than in JJA. The relative contribution of P to R is most important, followed by ET, with SM contributing minimally. The study emphasizes the need for further research to understand the sources of uncertainty in hydrological projections and improve the accuracy of future water condition projections.This study investigates the projected changes in runoff (R), precipitation (P), evapotranspiration (ET), and soil moisture (SM) under CMIP5 and CMIP6 scenarios, quantifying uncertainties at annual and seasonal scales. The results show that all four variables increase over most of the global land under a high emissions scenario (SSP5-8.5) during 2080–99, with 72%, 81%, 82%, and 66% of global land area showing increased R, P, ET, and SM, respectively. Model uncertainty dominates the total projected uncertainties, with 76% (R), 73% (P), 89% (ET), and 95% (SM) of uncertainties in the 2090s. Internal variability decreases over time, while scenario uncertainty increases. Low-latitude regions have the greatest uncertainty in hydrological projections. The uncertainty in P contributes the most to R uncertainty, with 93% at annual scale. CMIP5 and CMIP6 models show similar performance in terms of hydrological changes and uncertainty composition. The study provides a theoretical basis for improving hydrological components in global climate models. The results highlight the importance of quantifying uncertainty sources in hydrological projections to improve the reliability of climate models and future water condition projections. The study also compares CMIP5 and CMIP6 results, finding that CMIP6 has higher model agreement in some regions but larger uncertainties in long-term projections. The contribution of scenario uncertainty is higher in DJF than in JJA. The relative contribution of P to R is most important, followed by ET, with SM contributing minimally. The study emphasizes the need for further research to understand the sources of uncertainty in hydrological projections and improve the accuracy of future water condition projections.