This report documents the development of a computer model to simulate steady-state (long-term average) flow of ground water in the vicinity of Mirror Lake, which lies at the eastern end of the Hubbard Brook valley in central New Hampshire. The 10-km2 study area includes Mirror Lake, the three streams that flow into Mirror Lake, Leeman's Brook, Paradise Brook, and parts of Hubbard Brook and the Pemigewasset River. The topography of the area is characterized by steep hillsides and relatively flat valleys. Major hydrogeologic units include glacial deposits, composed of till containing pockets of sand and gravel, and fractured crystalline bedrock, composed of schist intruded by granite, pegmatite, and lamprophyre. Ground water occurs in both the glacial deposits and bedrock. Precipitation and snowmelt infiltrate to the water table on the hillsides, flow downslope through the saturated glacial deposits and fractured bedrock, and discharge to streams and to Mirror Lake. The model domain includes the glacial deposits, the uppermost 150m of bedrock, Mirror Lake, the layer of organic sediments on the lake bottom, and streams and rivers within the study area. A streamflow routing package was included in the model to simulate baseflow in streams and interaction between streams and ground water. Recharge from precipitation is assumed to be areally uniform, and riparian evapotranspiration along stream banks is assumed negligible. The spatial distribution of hydraulic conductivity is represented by dividing the model domain into several zones, each having uniform hydraulic properties. Local variations in recharge and hydraulic conductivities are ignored; therefore, the simulation results characterize the general ground-water system, not local details of ground-water movement. The model was calibrated using a nonlinear regression method to match hydraulic heads measured in piezometers and wells, and baseflow in three inlet streams to Mirror Lake. Model calibration indicates that recharge from precipitation to the water table is 26 to 28 cm/year. Hydraulic conductivities are 1.7 x 10-6 to 2.7 x 10-6 m/s for glacial deposits, about 3 x 10-7 m/s for bedrock beneath lower hillsides and valleys, and about 6x10-8 m/s for bedrock beneath upper hillsides and hilltops. Analysis of parameter uncertainty indicates that the above values are well constrained, at least within the context of regression analysis. In the regression, several attributes of the ground-water flow model are assumed perfectly known. The hydraulic conductivity for bedrock beneath upper hillsides and hilltops was determined from few data, and additional data are needed to further confirm this result. Model fit was not improved by introducing a 10-to-1 ration of horizontal-to-vertical anisotropy in the hydraulic conductivity of the glacial deposits, or by varying hydraulic conductivity with depth in the modeled part (uppermost 150m) of the bedrock. The calibrated model was used to delineate the Mirror Lake ground-water basin, defined as the volumes of subsurface through which ground water flows from the water table to Mirror Lake or its inlet streams. Results indicate that Mirror Lake and its inlet streams drain an area of ground-water recharge that is about 1.5 times the area of the surface-water basin. The ground-water basin extends far up the hillside on the northwestern part of the study area. Ground water from this area flows at depth under Norris Brook to discharge into Mirror Lake or its inlet streams. As a result, the Mirror Lake ground-water basin extends beneath the adjacent ground-water basin that drains into Norris Brook. Model simulation indicates that approximately 300,000 m3/year of precipitation recharges the Mirror Lake ground-water basin. About half the recharge enters the basin in areas where the simulated water table lies in glacial deposits; the other half enters the basin in areas where the simulated water table lies in be