Areas contributing recharge and sources of water to a production well field in the Village of Harrisville and to a production well field in the Town of Richmond were delineated on the basis of calibrated, steady-state ground-water-flow models representing average hydrologic conditions. The study sites represent contrasting glacial valley-fill settings. The area contributing recharge to a well is defined as the surface area where water recharges the ground water and then flows toward and discharges to the well. In Harrisville, the production well field is composed of three wells in a narrow, approximately 0.5-mile-wide, valley-fill setting on opposite sides of Batty Brook, a small intermittent stream that drains 0.64 square mile at its confluence with the Clear River. Glacial stratified deposits are generally less areally extensive than previously published. The production wells are screened in a thin (30 feet) but transmissive aquifer. Paired measurements of ground-water and surface-water levels indicated that the direction of flow between the brook and the aquifer was generally downward during pumping conditions. Long-term mean annual streamflow from two streams upgradient of the well field totaled 0.72 cubic feet per second. The simulated area contributing recharge for the 2005 average well-field withdrawal rate of 224 gallons per minute extended upgradient to ground-water divides in upland areas and encompassed 0.17 square mile. The well field derived 62 percent of pumped water from intercepted ground water and 38 percent from infiltrated stream water from the Batty Brook watershed. For the maximum simulated well-field withdrawal of 600 gallons per minute, the area contributing recharge expanded to 0.44 square mile to intercept additional ground water and infiltration of stream water; the percentage of water derived from surface water, however, was the same as for the average pumping rate. Because of the small size of Batty Brook watershed, most of the precipitation recharge in the watershed was withdrawn by the well field at the maximum rate either by intercepted ground water or indirectly by infiltrated stream water. Because the production wells are screened in a thin and transmissive aquifer in a small watershed, simulated ground-water traveltimes from recharge locations to the discharging wells were relatively short: 93 percent of the traveltimes were 10 years or less. In Richmond, the production well field is composed of two wells adjacent to and east of the Wood River in a moderately broad, approximately 1.2-mile-wide, valley-fill setting. The wells are screened in a transmissive aquifer with saturated thickness greater than 60 feet. Streamflow measurements in Baker Brook, a tributary to the Wood River 0.4 mile north of the well-field site, indicated that natural net loss of streamflow between the upland-valley contact and a downstream site was 0.12 cubic feet per second under average hydrologic conditions. Simulated areas contributing recharge for the maximum well-field pumping rate of 675 gallons per minute and for one-half the maximum rate extended northeastward from the well field to ground-water divides in upland areas. The area contributing recharge also included a remote, isolated area on the opposite side of the Wood River from the well field. The model simulation indicated that the well field did not derive any of its water from the Wood River because of the large watershed and associated quantity of ground water available for capture by the well field. The area contributing recharge for one-half the maximum rate was 0.31 square mile and the primary source of water to the well field was direct precipitation recharge. Fifteen percent of the water withdrawn from the production wells, however, was obtained from Baker Brook, indicating the importance of even small, distant tributary streams to the contributing area to a well. The area contributing recharge on the opposite side of the Wood River is