The stratified-drift aquifer in the 3,000-ft (feet)-wide and 100-ft-deep buried valley of Killbuck Creek near Wooster in northeastern Ohio was studied. The stratified drift with adjacent sandstone and shale bedrock produce a system of ground-water flow representative of the western part of the glaciated north-eastern United States. The stratified-drift aquifer is an excellent source of water for municipal and industrial wells. The aquifer is recharged locally by water from precipitation on the valley floor and uplands, by infiltration from streams, and by lateral flow to the valley from the uplands. As a result, the aquifer is vulnerable to surface or subsurface spills of contaminants in the valley or the adjacent uplands. Quality of water in the stratified drift is affected by influx of water from bedrock lateral to or beneath the valley. This influx is controlled, in part, by the pumping stress placed on the stratified-drift aquifer. Hydrogeologic and aqueous-geochemical data were analyzed to establish the framework necessary for stead-state and transient simulations of ground-water flow in stratified drift and bedrock with a three-layer ground-water-flow model. A new model routine, the Variable-Recharge procedure, was developed to simulate areal recharge and the contribution of the uplands to the drift system. This procedure allows for water applied to land surface to infiltrate or to be rejected. Rejected recharge and ground water discharged when the water table is at land surface form surface runoff-this excess upland water can be redirected as runoff to other parts of the model. Infiltration of streamwater, areal recharge to uplands and valley, and lateral subsurface flow from the uplands to the valley are sources of water to the stratufued0druft aquifer. Water is removed from the stratified-drift aquifer at Wooster primarily by production wells pumping at a rate of approximately 8.5 ft3/s (cubic feet per second). The ground-water budget resulting from two types of simulations of ground-water flow in this study indicates the primary sources of water to the wells are recharge at or near land surface and lateral subsurface flow from the shale and sandstone bedrock. Components of recharge at land surface include induced infiltration from streams, precipitation on the valley floor, and infiltration of unchanneled upland runoff that reaches the valley floor. The steady-state simulation was designed to represent conditions during the fall of 1984. The transient simulation was designed to represent an 11-day snowmelt event, 23 February to 5 March 1985, that caused water levels to rise significantly throughout the valley. Areal recharge to the valley and flow from the uplands to the valley were determined through the Variable-Recharge procedure. The total steady-state recharge to the valley was 12.5 ft3/s. Upland sources, areal valley recharge, and induced infiltration from Killnuck Creek accounted for 63, 23, and 8 percent, respectively, of the valley recharge. An analysis of the simulated vertical flow to the buried stratified drift through surficial slit, clay, and fine sand indicates that about 75 percent of the total recharge to the buried deposits is the sum of areally extensive, relatively small flows less than about 0.01 ft? /s per model node), whereas about 25 percent of the recharge results from a really restricted, relatively large flows (greater than about 0.01 ft? /s per model node). The large-magnitude flows are located primarily beneath Clear and Little Killbuck Creeks where seepage provides abundant recharge and the surficial sediments grade into coarser alluvial-fan deposits. Chemical and isotopic studies of ground water and streamwater combined with measurements of stream infiltration provide independent support for the conclusions derived from computer simulation of ground-water flow. In addition, the chemical and isotopic studies helped quantity the rate and pathways of infiltrating water from