Nutrient loads from ground-water discharge were studied in Broad Brook Basin, a 15.8-square mile basin in north-central Connecticut, dominated by agricultural activity. Loads were calculated, along with the travel times of ground water from recharge to discharge areas, to estimate the time required for the effects of Best Management Practices (BMPs) to be observed. Most concentrations of nitrogen and phosphorus in Broad Brook exceeded U.S. Environmental Protection Agency Ecoregion XIV nutrient criteria for streams. During the study period (1993-2004), annual loads of nitrogen from Broad Brook Basin ranged from 117,000 to 270,000 pounds (lb), and yields were about 10 times larger than those from forested basins in Connecticut. Ground-water discharge from the aquifer to the streams (base flow) during the study period was estimated with hydrograph separation and accounted for 82 percent of the total runoff from the basin. Nitrate nitrogen in base flow averaged 71 percent of the annual load of total nitrogen discharged from the basin, indicating that the largest source of nitrogen was likely from ground-water discharge. Annual loads of total phosphorus from the basin ranged from 2,330 to 14,400 lb, and yields were about five times higher than those from forested basins in Connecticut. Dissolved phosphorus averaged about 71 percent of the total phosphorus load, and ground-water discharge accounted for only as much as 40 percent of the annual load of dissolved phosphorus; therefore, phosphorus loads are dominated by stormwater-runoff events. Ground-water samples collected from 11 wells in the basin contained elevated concentrations of nitrite plus nitrate nitrogen. Dissolved gas analyses indicated that little denitrification was occurring in the aquifer. Apparent ages of the ground-water samples ranged from greater than 2 to more than 50 years based on sulfur hexafluoride, tritium, and tritium/helium-3 analyses. A three-dimensional ground-water-flow model was used in conjunction with a particle-tracking program to determine advective travel times to streams from all subareas in the basin. The model simulations indicated that about half the discharge to Broad Brook and its tributaries was recharged more than 10 years ago, and that about 8 percent of the discharge was recharged prior to 1960. The effects of changes in nitrate nitrogen loading at the water table were evaluated by applying new loading rates from urban and agricultural land and the simulated advective ground-water travel times. Five scenarios were tested: reducing estimated nitrate nitrogen concentrations in recharging ground water under urban and agricultural land areas to concentrations in forested areas; reducing estimated nitrate nitrogen concentrations under urban and agricultural land areas to the U.S. Environmental Protection Agency recommended nutrient criteria for streams; and reducing estimated nitrate nitrogen concentrations under urban and agricultural land areas by 50 percent, 10 percent, and 5 percent. Under the first two scenarios, the base-flow load of nitrate nitrogen could be reduced by 25 percent in slightly more than 5 years, although the reduction required by these scenarios is likely unrealistic. A 25-percent reduction in base-flow load of nitrate nitrogen could be achieved in about 10 years under the third scenario (where concentrations from urban and agricultural areas were reduced by 50 percent). Under this scenario, a 46-percent reduction could be achieved in about 60 years. The scenarios indicate that in this basin, and in other similar basins in Connecticut underlain by extensive glacial stratified deposits, there can be a substantial time lag between activities at the land surface and effects on the quality of water discharged to streams from ground water. This finding may be important in the expectations for water-quality improvements from land-use changes or BMPs.