A three-dimensional finite-difference digital model was used to simulate ground-water flow in the 25,000-square-mile aquifer system of the North Carolina Coastal Plain. The model was developed from a hydrogeologic framework that is based on an alternating sequence of 10 aquifers and 9 confining units, which make up a seaward-thickening wedge of sediments that form the Coastal Plain aquifer system in the State of North Carolina. The model was calibrated by comparing observed and simulated water levels. The model calibration was achieved by adjusting model parameters, primarily leakance of confining units and transmissivity of aquifers, until differences between observed and simulated water levels were within acceptable limits, generally within 15 feet. The maximum transmissivity of an individual aquifer in the calibrated model is 200,000 feet squared per day in a part of the Castle Hayne aquifer, which consists predominantly of limestone. The maximum value for simulated vertical hydraulic conductivity in a confining unit was 2.5 feet per day, in a part of the confining unit overlying the upper Cape Fear aquifer. The minimum value was 4.1x10-6 feet per day, in part of the confining unit overlying the lower Cape Fear aquifer. Analysis indicated the model is highly sensitive to changes in transmissivity and leakance near pumping centers; away from pumping centers, the model is only slightly sensitive to changes in transmissivity but is moderately sensitive to changes in leakance. Recharge from precipitation to the surficial aquifer ranges from about 12 inches per year in areas having clay at the surface to about 20 inches per year in areas having sand at the surface. Most of this recharge moves laterally to streams, and only about 1 inch per year moves downward to the confined parts of the aquifer system. Under predevelopment conditions, the confined aquifers were generally recharged in updip interstream areas and discharged through streambeds and in downdip coastward areas. Hydrologic analysis of the flow system using the calibrated model indicated that, because of ground-water withdrawals, areas of ground-water recharge have expanded and encroached upon some major stream valleys and into coastal area. Simulations of pumping conditions indicate that by 1980 large parts of the former coastal discharge areas had become areas of potential or actual recharge. Declines of ground-water level, which are the result of water taken from storage, are extensive in some areas and minimal in others. Hydraulic head declines of more than 135 feet have occurred in the northern Coastal Plain since 1940 primarily due to withdrawals in the Franklin area in Virginia. Declines of ground-water levels greater than 110 feet have occurred in aquifers in the central Coastal Plain due to combined effects of pumpage for public and industrial water supplies. Water-level declines exceeding 100 feet have occurred in the Beaufort County area because of withdrawals for a mining operation and water supplies for a chemical plant. Head declines have been less than 10 feet in the shallow surficial and Yorktown aquifers and in the updip parts of the major confined aquifers distant from areas of major withdrawals. In 1980, contribution from aquifer storage was 14 cubic feet per second, which is about 4.8 percent of pumpage and about 0.05 percent of ground-water recharge. A water-budget analysis using the model simulations indicates that much of the water removed from the ground-water system by pumping ultimately is made up by a reduction in water leaving the aquifer system, which discharges to streams as base flow. The reduction in stream base flow was 294 cubic feet per second in 1980 and represents about 1.1 percent of the ground-water recharge. The net reduction to streamflow is not large, however, because most pumped ground water is eventually discharged to streams. In places, such as at rock quarries in Onslow and Craven Counties, water is lost from st