The sensitivity of annual runoff to changes in two main climate variables, precipitation (P) and potential evapotranspiration (PET), was explored using a regression- based model and a soil water balance model (SWB-Model) in both a semi-arid and humid watershed in California. The SWB-Model was formulated using the top-down method of model development. A simple, “1-bucket” model with parameters for soil depth, vegetation cover, and evaporation was found to sufficiently predict annual runoff in the semi-arid catchment, while subsurface flows were required for the humid catchment. The SWB-Model was then used to test climate change scenarios, and the results were compared to sensitivity factors produced by the regression-based model. The SWB-Model produced similar estimates of the sensitivity of runoff to changes in climate to the regression-based model when there was no multicollinearity between P and PET; however, when P and PET were correlated, the sensitivity factors produced by the regression-based model were unstable. Overall, the models indicated that the semi-arid catchment was up to four times more sensitive to both changes in P and PET than the humid catchment. This suggests that water- limited systems have a stronger watershed response to climate variability than energy-limited systems.