In water-stressed, semiarid urban environments, connections between impervious surfaces and drainage networks may strongly impact the water use and ecosystem productivity of neighboring vegetated areas. We use an ecohydrologic model, the Regional Hydro-Ecological Simulation System (RHESSys), to quantify the sensitivity of vegetation water use and net primary productivity (NPP) to fine-scale impervious surface connectivity. We develop a set of very fine-scale (2 m2) scenarios that vary both the percentage of impervious surface and fraction of this impervious surface with direct hydrologic connections to urban drainage systems for a small hillslope. When driven by Mediterranean climate forcing, model estimates suggest that total vegetation water use declines with increasing impervious area. However, when impervious area is hydrologically disconnected from the urban drainage network, declines in water and carbon fluxes with decreased vegetated area can be partially, or in some cases even completely, offset by increased transpiration and NPP in the remaining vegetation. Relative increases in water use and NPP of remaining vegetation are much greater for deeply rooted shrubs and trees and negligible for shallow rooted grasses. We extrapolate our findings to the catchment scale by developing a first-order approximation of fine-scale impervious connection impacts on aggregate watershed water and carbon flux estimates. Our approach offers a computationally and data-efficient method for estimating the impact of impervious area connectivity on these ecohydrologic fluxes. For our only partially urbanized Santa Barbara watershed, estimates of water use and NPP that account for fine-scale impervious connection differed by more than 10% from those that did not.