Uncertainty in Land Surface Model (LSM) simulations can arise from various sources, including model structure, parameterization, and input data. However, there is a lack of understanding of how forcing selection and parametric uncertainties can influence LSMs simulated key terrestrial water budget components. To address this research need, we use the Community Land Model version 5 (CLM5), the land component of the Community Earth System Model, at a spatial scale of 12-km to simulate the land surface and subsurface processes for 464 watersheds, selected from the Catchment Attributes for Large-Sample Studies (CAMELS) dataset to be representative of physiographic and climatic gradients across the conterminous United States (CONUS). For each watershed, CLM5 experiments are driven by five most commonly used, gridded forcing datasets: NLDAS-2, Livneh, PRISM, Daymet, and WRF-based dynamical downscaling of ERA5 reanalysis data, in combination with a large ensemble (> 1300) of key CLM5 hydrologic parameters to represent parametric uncertainty. We diagnose the contribution of forcing selection to parametric uncertainty for eight hydrologic signatures including extreme events (drought and flood) and seasonal water balance. Our results suggest that uncertainty in lower streamflow quantiles is dominated by parametric uncertainty, while forcing uncertainty contributes more to higher streamflow quantiles. Our analysis also demonstrates the significant joint effects of forcing selection and parametric uncertainty on CLM5-based analyses of major regional drought and flood events using Texas and California examples. Overall the results from this study highlights the need to more fully understand and incorporate forcing selection and parametric uncertainties effects in LSM analyses particularly for hazard and risk assessments addressing hydrologic extremes.