The U.S. Geological Survey, in cooperation with the Town of Framingham, Massachusetts, has investigated the potential of proposed groundwater withdrawals at the Birch Road well site to affect nearby surface water bodies and wetlands, including Lake Cochituate, the Sudbury River, and the Great Meadows National Wildlife Refuge in east-central Massachusetts. In 2012, the U.S. Geological Survey developed a Phase 1 numerical groundwater model of a complex glacialsediment aquifer to synthesize hydrogeologic information and simulate potential future pumping scenarios. The model was developed with MODFLOW-NWT, an updated version of a standard USGS numerical groundwater flow modeling program that improves solution of unconfined groundwater flow problems. The groundwater model and investigations of the aquifer improved understanding of groundwater–surface-water interaction and the effects of groundwater withdrawals on surface-water bodies and wetlands in the study area. The initial work also revealed a need for additional information and model refinements to better understand this complex aquifer system. In this second phase of the study, the original groundwater flow model was revised to improve representation of groundwater and surface-water hydrology, stabilize the model, and reduce model error. The model was simplified by reducing the number of layers from 5 to 3 and adding the MODFLOW lake package (LAK) to simulate Lake Cochituate and Pod Meadow Pond and better represent interaction between the lakes and the aquifer. Model revisions improved stability and shortened run times, allowing use of automated parameter estimation software (PEST) to further refine the model hydraulic parameters and reduce simulation errors. Model simulations indicate that under average baseflow conditions, the Birch Road wells have a small effect on flow in the Sudbury River during most months, even at the maximum pumping rate of 4.9 ft3/s (3.17 Mgal/d). Maximum percent streamflow depletion in the Sudbury River caused by simulated pumping takes place during simulated drought conditions, when streamflow decreased by as much as 21 percent under maximum continuous pumping. Simulations also indicate that groundwater withdrawals at the Birch Road site could be managed so that adverse streamflow impacts are substantially ameliorated. Under the most ecologically conservative simulated drought conditions, simulated streamflow depletion was reduced from 21 percent to 3 percent by pumping at the maximum rate for 6 months rather than for 12 months. Simulations that return 10 percent of the Birch Road well withdrawals to Pod Meadow Pond indicate a modest reduction in the Sudbury River streamflow depletion and provide a larger percentage increase to streamflow just downstream of the pond. The groundwater model also indicates that well locations can have a large effect on the sustainable pumping rate and so should be chosen carefully. The model provides a tool for evaluating alternative pumping rates and schedules not included in this analysis.