A 27-day streamflow augmentation test was conducted in December 1979 at Fosters Brook, near the south shore of Long Island, to investigate the hydraulic feasibility of pumping ground water to supply flow to an ephemeral stream during dry periods. Measurements of soil moisture in the unsaturated zone beneath the streambed indicate that infiltration rate and soil-moisture content are interrelated. Initial infiltration was measured with a neutron logger; the wetting front traversed the unsaturated zone at an average of 11.2 inches per hour and reached the water table in 5.5 hours. Soil moisture in the unsaturated zone ranged from 20 percent at the start of the test to nearly 41 percent, nearly the saturation point, 20 days later. Stream discharge was measured at four sites along the stream channel, and the augmentation rate was monitored continuously at the starting point. Infiltration rates increased steadily in all reaches during the first 12 days of the test, but from the 12th to the 20th day, when discharge was increased by 50 percent, infiltration rates decreased along the two upstream reaches but continued to increase along the three downstream reaches. Infiltration rates remained constant from days 20 through 26. During the first 24 hours of the test, the stream reached a maximum length of 2,050 feet, but after 13 days, it had shortened to 1,300 feet as a result of seepage losses. The relationship between discharge and stream length was linear within the range of discharge investigated (0.54-1.63 cubic feet per second). Ground-water levels rose in response to flow augmentation and reached a maximum rise of about 6.5 feet in a well situated 14 feet from the center of the streambed and 225 feet downstream from the start of the flow. Measured water-level response was compared to levels predicted by a one-dimensional analytical model and a three-dimensional mathematical model; results indicate that ground-water response is determined principally by streambed characteristics and soil-moisture content in the unsaturated zone. Variations in water temperature and in streambed composition had significant effects upon infiltration rates. Changes in water temperature, amount of vegetation, soil-moisture content, and stream stage, combined with local variations in streambed permeability and aquifer conductivity, make accurate prediction of seepage rates virtually impossible at present. Data from this study suggest that site-specific investigations are necessary wherever streamflow augmentation is planned.