Eastern hemlock (Tsuga canadensis (L.) Carr.) is one of the principal riparian and cove canopy species in the southern Appalachian Mountains. Throughout its range, eastern hemlock is facing potential widespread mortality from the hemlock woolly adelgid (HWA). If HWA-induced eastern hemlock mortality alters hydrologic function, land managers will be challenged to develop management strategies that restore function or mitigate impacts. To estimate the impact that the loss of this forest species will have on the hydrologic budget, we quantified and modeled transpiration over a range of tree sizes and environmental conditions. We used heat dissipation probes, leaf-level gas-exchange measurements, allometric scaling, and time series modeling techniques to quantify whole-tree and leaf-level transpiration (E(L)) of eastern hemlock. We monitored trees ranging from 9.5 to 67.5 cm in diameter along a riparian corridor in western North Carolina, USA during 2004 and 2005. Maximum rates of daily tree water use varied by diameter and height, with large trees transpiring a maximum of 178-186 kg H2O x tree(-1) x d(-1). Values of E(L) could be predicted from current and lagged environmental variables. We forecasted eastern hemlock E(L) for inventoried stands and estimated a mean annual transpiration rate of 63.3 mm/yr for the hemlock component, with 50% being transpired in the winter and spring. In typical southern Appalachian stands, eastern hemlock mortality would thus reduce annual stand-level transpiration by approximately 10% and reduce winter and spring stand-level transpiration by approximately 30%. Eastern hemlock in the southern Appalachians has two distinct ecohydrological roles: an evergreen tree that maintains year-round transpiration rates and a riparian tree that has high transpiration rates in the spring. No other native evergreen in the southern Appalachians will likely fill the ecohydrological role of eastern hemlock if widespread mortality occurs. With the loss of this species, we predict persistent increases in discharge, decreases in the diurnal amplitude of streamflow, and increases in the width of the variable source area.