The U.S. Geological Survey, in cooperation with the Oklahoma Water Resources Board, began a three-year study of the High Plains aquifer in northwestern Oklahoma in 1996. The primary purpose of this study was to develop a ground-water flow model to provide the Water Board with the information it needs to manage the quantity of water withdrawn from the aquifer. The study area consists of about 7,100 square miles in Oklahoma and about 20,800 square miles in adjacent states to provide appropriate hydrologic boundaries for the flow model. The High Plains aquifer includes all sediments from the base of the Ogallala Formation to the potentiometric surface. The saturated thickness in Oklahoma ranges from more than 400 feet to less than 50 feet. Natural recharge to the aquifer from precipitation occurs throughout the area but is extremely variable. Dryland agricultural practices appear to enhance recharge from precipitation, and part of the water pumped for irrigation also recharges the aquifer. Natural discharge occurs as discharge to streams, evapotranspiration where the depth to water is shallow, and diffuse ground-water flow across the eastern boundary. Artificial discharge occurs as discharge to wells. Irrigation accounted for 96 percent of all use of water from the High Plains aquifer in the Oklahoma portion of the study area in 1992 and 93 percent in 1997. Total estimated water use in 1992 for the Oklahoma portion of the study area was 396,000 acre-feet and was about 3.2 million acre-feet for the entire study area. Since development of the aquifer, water levels have declined more than 100 feet in small areas of Texas County, Oklahoma, and more than 50 feet in areas of Cimarron County. Only a small area of Beaver County had declines of more than 10 feet, and Ellis County had rises of more than 10 feet. A flow model constructed using the MODFLOW computer code had 21,073 active cells in one layer and had a 6,000- foot grid in both the north-south and east-west directions. The model was used to simulate the period before major development of the aquifer and the period of development. The model was calibrated using observed conditions available as of 1998. The predevelopment-period model integrated data or estimates on the base of aquifer, hydraulic conductivity, streambed and drain conductances, and recharge from precipitation to calculate the predevelopment altitude of the water table, discharge to the rivers and streams, and other discharges. Hydraulic conductivity, recharge, and streambed conductance were varied during calibration so that the model produced a reasonable representation of the observed water table altitude and the estimated discharge to streams. Hydraulic conductivity was reduced in the area of salt dissolution in underlying Permianage rocks. Recharge from precipitation was estimated to be 4.0 percent of precipitation in greater recharge zones and 0.37 percent in lesser recharge zones. Within Oklahoma, the mean difference between water levels simulated by the model and measured water levels at 86 observation points is -2.8 feet, the mean absolute difference is 44.1 feet, and the root mean square difference is 52.0 feet. The simulated discharge is much larger than the estimated discharge for the Beaver River, is somewhat larger for Cimarron River and Wolf Creek, and is about the same for Crooked Creek. The development-period model added specific yield, pumpage, and recharge due to irrigation and dryland cultivation to simulate the period 1946 through 1997. During calibration, estimated specific yield was reduced by 15 percent in Oklahoma east of the Cimarron-Texas County line. Simulated recharge due to irrigation ranges from 24 percent for the 1940s and 1950s to 2 percent for the 1990s. Estimated recharge due to dryland cultivation is about 3.9 percent of precipitation. The mean difference between the simulated and observed waterlevel changes from predevelopment to 1998 at 162 observation points in