The water cycle is an important component of the earth system and it plays a key role in many facets of society, including energy production, agriculture, and human health and safety. Earth system models have been an important tool for deepening our understanding and testing hypotheses related to the water cycle. Continued advances in computing power have enabled earth system models to resolve features important to the water cycle that lower resolution models often fail to capture. No single number can represent all the relevant processes and scales related to the water cycle, and as a result, many metrics have been proposed to quantitatively assess individual features of the water cycle. In this study, a representative set of metrics related to the water cycle have been selected to evaluate the Energy Exascale Earth System Model (E3SM) at low- and high-resolution – as established by the HighResMIP protocol. Focusing on the conterminous United States (CONUS), the performance of E3SM is evaluated at the watershed scale for metrics including the spatial distribution of rainfall, its seasonality, diurnal cycle, frequency of occurrence, and extremes, as well as several surface component terms: snowpack, runoff, and streamflow. Results reveal that different metrics are sensitive to resolution in different watersheds, such that no all-inclusive statement can be made about the CONUS water cycle response to increasing resolution. The physical pathways giving rise to some of these differences are detailed for a subset of watersheds.