Areas contributing water to the Dry Brook public-supply well in South Hadley, Massachusetts, were delineated with a numerical ground-water-flow model that is based on geologic and hydrologic information for the confined sand and gravel aquifer pumped by the supply well. The study area is along the Connecticut River in central Massachusetts, about 12 miles north of Springfield, Massachusetts. Geologic units in the study area consist of Mesozoic-aged sedimentary and igneous bedrock, late-Pleistocene glaciolacustrine sediments, and recent alluvial deposits of the Connecticut River flood plain. Dry Brook Hill, immediately south of the supply well, is a large subaqueous lacustrine fan and delta formed during the last glacial retreat by sediment deposition into glacial Lake Hitchcock from a meltwater tunnel that was likely near where the Connecticut River cuts through the Holyoke Range. The sediments that compose the aquifer grade from very coarse sand and gravel along the northern flank of the hill, to medium sands in the body of the hill, and to finer-grained sediments along the southern flank of the hill. The interbedded and overlapping fine-grained lacustrine sediments associated with Dry Brook Hill include varved silt and clay deposits. These fine-grained sediments form a confining bed above the coarse-grained aquifer at the supply well and partially extend under the Connecticut River adjacent to the supply well. Ground-water flow in the aquifer supplying water to Dry Brook well was simulated with the U.S. Geological Survey ground-water-flow modeling code MODFLOW. The Dry Brook aquifer model was calibrated to drawdown data collected from 8 observation wells during an aquifer test conducted by pumping the supply well for 10 days at a rate of 122.2 cubic feet per minute (ft3/min; 914 gallons per minute) and to water levels collected from observation wells across the study area. Generally, the largest hydraulic conductivity values used in the model were in the sand and gravel aquifer near the Dry Brook well, which is consistent with the geologic information. Results of aquifer-test simulation indicated that spatially variable aquifer hydraulic properties and boundary conditions affected heads and ground-water flow near the well. A comparison and analysis of water-level fluctuations in study area wells to fluctuations in the Connecticut River indicated a hydraulic connection of the aquifer with the river, which is also consistent with geologic information. Simulated ground-water levels indicated that most ground water in the study area flowed toward and discharged to the Connecticut River and the Dry Brook well. Small amounts of ground water also discharged to smaller streams (Dry Brook and Bachelor Brook) in the study area. Areas contributing water to the well were delineated with the MODPATH particle-tracking routine. Results of the contributing-area analysis indicated that the greatest sources of water to the well were recharge in the Dry Brook Hill area and infiltration of Connecticut River water in an area beyond the extent of the confining bed where the aquifer is in hydraulic connection with the river. The amount of water entering the Dry Brook well from recharge dominated at a lower pumping rate (40.0 ft3/min); about 90 percent of the pumped water originated from recharge and boundary flow, and infiltration from the Connecticut River supplied the remaining 10 percent. At a high pumping rate (122.2 ft3/min), however, about half of the water pumped from the Dry Brook well originated from recharge and boundary flow (49 percent), and half originated from infiltration of water from the Connecticut River (51 percent). Results of a sensitivity analysis of the extent of areas contributing water to the Dry Brook well when pumped at 122.2 ft3/min indicated that the size of these areas did not substantially change when aquifer properties were varied. In contrast, however, the size of these areas changed most when the recharge