A regional MODFLOW ground-water flow model of parts of coastal Georgia, Florida, and South Carolina was used to evaluate the effects of current and hypothetical groundwater withdrawal, and the relative effects of pumping in specific areas on ground-water flow in the Upper Floridan aquifer near Fort Stewart and Hunter Army Airfield (HAAF), coastal Georgia. Simulation results for four steady-state pumping scenarios were compared to each other and to a Base Case condition. The Base Case represents year 2000 pumping rates throughout the model area, with the exception that permitted annual average pumping rates for the year 2005 were used for 26 production wells at Fort Stewart and HAAF. The four pumping scenarios focused on pumping increases at HAAF resulting from projected future demands and additional personnel stationed at the facility and on reductions in pumping at Fort Stewart. Scenarios A and B simulate 1- and 2-million-gallon-perday (Mgal/d) increases, respectively, at HAAF. Simulated water-level change maps for these scenarios indicate an area of influence that extends into parts of Bryan, Bulloch, Chatham, Effingham, and Liberty Counties, Ga., and Beaufort and Jasper Counties, S.C., with maximum drawdowns from 0.5 to 4 feet (ft) for scenario A and 1 to 8 ft for Scenario B. For scenarios C and D, increases in pumping at HAAF were offset by decreases in pumping at Fort Stewart. Scenario C represents a 1-Mgal/d increase at HAAF and a 1-Mgal/d decrease at Fort Stewart; simulated water-level changes range from 0.4 to -4 ft. Scenario D represents a 2-Mgal/d increase at HAAF and 2-Mgal/d decrease at Fort Stewart; simulated water-level changes range from 0.04 to -8 ft. The simulated water-level changes indicate an area of influence that extends into parts of Bryan, Bulloch, Chatham, Effingham, Liberty, and McIntosh Counties, Ga., and Jasper and Beaufort Counties, S.C. In general, decreasing pumping at Fort Stewart by an equivalent amount to pumping increases at HAAF reduced the magnitude and extent of drawdown resulting from the additional pumping. None of the scenarios resulted in large changes in the configuration of the simulated potentiometric surface and related ground-water flow directions. The scenarios simulated vary from the original model only by increasing pumpage less than 1 percent of the total calibrated model withdrawals. The changes in pumpage are located near the center of the original model area. Thus, the scenarios described in this report are considered to be reasonable with no less uncertainty than the original calibrated model.