In the face of climate change droughts are predicted to become more persistent, further intensifying the need for accurately predicting the timing and magnitude of summer streamflow in rivers. In order to determine the sensitivity of a watershed to drought, there is a need to describe what these drought conditions will look like and to quantify how severe the effects on the rivers and aquifers will be. Discrepancies between model predictions and observed streamflow motivate us to improve our current understanding of watershed responses to drought in order to better characterize it. This dissertation attempts to identify issues surrounding the current understanding of watershed responses to drought. First, we introduce a methodology that can greatly expand the number of watersheds analyzed using recession analysis with water height instead of discharge. Using water height for recession analysis instead of discharge reduces the fieldwork associated with data collection and thus with the same resources more watersheds can be characterized using water height alone. Second, we focused on quantifying the impacts of different recession analysis methods on the parameter estimations and consequences for drought sensitivity interpretation. We conclude that analyzing the recessions collectively is flawed and that analysis should consider individual recessions in order to quantify watershed responses to different hydrological scenarios. Third, we examined how hillslope drainage pathways and residence time varies during a drought event and the associated effects of modeling the process in 2D vs 3D. In 3D, the hillslope drainage path rotates from stream perpendicular to stream parallel as the contributing driving gradient changes with the lowering water table. This results in an increased path length and residence time that is not captured in a 2D model, ultimately changing the timing and magnitude of aquifer contributions to streamflow during drought. While this dissertation has developed a framework for better understanding and predicting streamflow during drought, there are still opportunities to improve the characterization by monitoring more small watersheds, describing the variability in individual recession at low flows, and gaining a deeper understanding of drainage timescale. Knowing how streams and aquifers will respond to drought in future climate scenarios has great implications for water management as it may enable us to identify watersheds that are sensitive or resilient to future drought.