A digital model was developed to simulate steady-state ground-water flow in a 42,155-square-mile area of coastal Georgia and adjacent parts of South Carolina and Florida. The model was developed to (1) understand and refine the conceptual model of regional ground-water flow, (2) serve as a framework for the development of digital subregional ground-water flow and solute-transport models, and (3) serve as a tool for future evaluations of hypothetical pumping scenarios used to facilitate water management in the coastal area. Single-density ground-water flow was simulated using the U.S. Geological Survey finite-difference code MODFLOW-2000 for mean-annual conditions during predevelopment (pre?1900) and the years 1980 and 2000. The model comprises seven layers: the surficial aquifer system, the Brunswick aquifer system, the Upper Floridan aquifer, the Lower Floridan aquifer, and the intervening confining units. A combination of boundary conditions was applied, including a general-head boundary condition on the top active cells of the model and a time-variable fixed-head boundary condition along part of the southern lateral boundary. Simulated heads for 1980 and 2000 conditions indicate a good match to observed values, based on a plus-or-minus 10-foot (ft) calibration target and calibration statistics. The root-mean square of residual water levels for the Upper Floridan aquifer was 13.0 ft for the 1980 calibration and 9.94 ft for the 2000 calibration. Some spatial patterns of residuals were indicated for the 1980 and 2000 simulations, and are likely a result of model-grid cell size and insufficiently detailed hydraulic-property and pumpage data in some areas. Simulated potentiometric surfaces for predevelopment, 1980, and 2000 conditions all show major flow system features that are indicated by estimated peotentiometric maps. During 1980?2000, simulated water levels at the centers of pumping at Savannah and Brunswick rose more than 20 ft and 8 ft, respectively, in response to decreased pumping. Simulated drawdown exceeded 10 ft in the Upper Floridan aquifer across much of the western half of the model area, with drawdown exceeding 20 ft along parts of the western, northern, and southern boundaries where irrigation pumping increased during this period. From predevelopment to 2000 conditions, the simulated water budget showed an increase in inflow from, and decrease in outflow to, the general-head boundaries, and a reversal from net seaward flow to net landward flow across the coastline. Simulated changes in recharge and discharge distribution from predevelopment to 2000 conditions showed an increase in extent and magnitude of net recharge cells in the northern part of the model area, and a decrease in discharge or change to recharge in cells containing major streams and beneath major pumping centers. The model is relatively sensitive to pumping and the controlling head at the fixed-head boundary and less sensitive to the distribution of aquifer properties in general. Model limitations include: (1) its spatial scale and discretization, (2) the extent to which data are available to physically define the flow system, (3) the type of boundary conditions and controlling parameters used, (4) uncertainty in the distribution of pumping, and (5) uncertainty in field-scale hydraulic properties. The model could be improved with more accurate estimates of ground-water pumpage and better characterization of recharge and discharge.