The demand for water in Thurston County has increased steadily in recent years because of a rapid growth in population. Surface-water resources in the county have been fully appropriated for many years and Thurston County now relies entirely on ground water for new supplies of water. Thurston County is underlain by up to 2,000 feet of unconsolidated glacial and non-glacial Quaternary sediments which overlie consolidated rocks of Tertiary age. Six geohydrologic units have been identified within the unconsolidated sediments. Between 1988 and 1990, median water levels rose 0.6 to 1.9 feet in all geohydrologic units except bedrock, in which they declined 1.4 feet. Greater wet-season precipitation in 1990 (43 inches) than in 1988 (26 inches) was the probable cause of the higher 1990 water levels. Ground-water flow in the unconsolidated sediments underlying Thurston County was simulated with a computerized numerical model (MODFLOW). The model was constructed to simulate 1988 ground-water conditions as steady state. Simulated inflow to the model area from precipitation and secondary recharge was 620,000 acre-feet per year (93 percent), leakage from streams and lakes was 38,000 acre-ft/yr (6 percent), and ground water entering the model along the Chehalis River valley was 5,800 acre-ft/yr (1 percent). Simulated outflow from the model was primarily leakage to streams, springs, lakes, and seepage faces (500,000 acre-ft/yr or 75 percent of the total outflow). Submarine seepage to Puget Sound was simulated to be 88,000 acre-ft/yr (13 percent). Simulated ground-water discharge along the Chehalis River valley was simulated to be 12,000 acreft/yr (2 percent). Simulated withdrawals by wells for all purposes was 62,000 acre-ft/yr (9 percent). The numerical model was used to simulate the possible effects of increasing ground-water withdrawals by 23,000 acre-ft/yr above the 1988 rate of withdrawal. The model indicated that the increased withdrawals would come from reduced discharge to springs, seepage faces, and offshore (total of 51 percent of increased pumping) and decreased flow to rivers (46 percent). About 3 percent would come from increased leakage from rivers. Water levels would decline more than 1 foot over most of the model area, more than 10 feet over some areas, and would be at a maximum of about 35 feet. Contributing areas for water discharging at McAllister and Abbott Springs and to pumping centers near Tumwater and Lacey were estimated using a particle-tracking post-processing computer code (MODPATH) and a MODFLOW model calibrated to steady-state (1988) conditions. Water discharging at McAllister and Abbot Springs was determined to come from water entering the ground-water system at the water table in an area of about 20 square miles (mi2) to the west and south of the springs. This water is estimated to come from recharge (both precipitation and secondary) and from leakage from Lake St. Clair and several other surface-water bodies. Southeast of Lacey, about 3,800 acre-ft of ground water were pumped from five municipal wells during 1988. The source of the pumped water was determined to be an area that covers about 1.1 mi2. The water was estimated to come from recharge (both precipitation and secondary) and leakage from surface-water bodies. Along the lower Deschutes River nearly 3,900 acre-ft/yr of ground water were pumped during 1988 from 15 wells for municipal and industrial use. The calculated source of this water was an area that covers about 1.3 mi2. Within the calculated contributing area the pumped ground water comes from recharge (both precipitation and secondary) and leakage from the Deschutes River and several other surface-water bodies.