Soil moisture is a key variable in determining water and energy fluxes between the land surface and the atmosphere, thus playing a critical role in driving local to continental-scale Earth system dynamics. Models involving soil moisture are also crucial in understanding how many coupled natural-human systems will respond to changing environmental conditions; examples include assessing drought or flood risk, simulating agricultural yields, and characterizing ecosystem services. Long-term projections of soil moisture are uncertain owing to several factors, including the representation of soil dynamics in land surface models and potential biases the simulation of future climate. Here, we performed a combined sensitivity analysis of the University of New Hampshire Water Balance Model (WBM) to understand which uncertain factors drive consequential changes in projections of future soil moisture. We focus on root-zone soil moisture in the central United States and analyze metrics of long-term change as well as indices of spatiotemporally connected extremes. We consider two sources of uncertainty: first, related to parameter choices within WBM, and second, related to high-resolution climate model projections. Our results paint a complex picture of interacting uncertainties across space, time, and metric of soil moisture change. Our framework also facilitates future uncertainty analyses on projections of crop yields and downstream effects on market price volatility.