Thesis viiProjections of surface runoff at large river watersheds is critical to inform water resources planner and policymakers. Although, Earth System Models (ESMs) provide runoff projections over the world, there is a large gap in projections of runoff over stakeholder-relevant spatiotemporal resolutions. In contrast, hydrological models provide runoffs at high spatial resolution but do not have the ability to incorporate feedbacks from atmospheric circulations for long term predictions. This study assesses the credibility of the state-of-the-art Earth System Models from CMIP6 generation to understand and project large river watershed runoffs. To assess the performance, statistical parameters (mean, variance, trends, MSE, NRMSE, CC, KGE, NSE, Bias) of Multi Model Ensemble (MME) from CMIP6 ESMs have been compared to reanalysis and observed runoffs for the largest 30 rivers in the world. Comparison was conducted at historical as well as future (2017-2100) periods under SSP 370 scenario. This study also estimates the number of people who will be impacted due to future runoff changes. Here we found that, the basin averaged runoff of ESMs MME has high bias and low relative variability in comparison to observations at seasonal and annual scale. But mean and trends are captured better in CMIP6 MMEs at climatological scale. Moreover, CMIP6 MME runoffs perform differently based on the factor of human interventions in river basins (managed vs unmanaged river basins). Analysis also shows that in the future, for SSP 370 scenario, mean, variance, trend will change (increase and decrease) based on river location and other climate factors. Furthermore, it was noted that, there is wide variability due to forcing, model response, or internal variability among models at each watershed basins which amplifies the uncertainties of projections. However, despite gaps in process understanding, as well as intrinsic variability, the projected changes in runoff at regional and climatological scales are significant enough to require re-evaluation of design curves, planning scenarios, and operations practices. vii viii