Skaneateles Lake, the fifth largest Finger Lake in NY, provides potable water to 160,000 residents of the greater Syracuse metropolitan area.  Agriculture plays an important role in the watershed.  The Skaneateles Lake Watershed Agriculture Program (SLWAP), initiated in 1994 as part of an agreement that allows water to be withdrawn from the lake unfiltered, has invested millions of dollars to improve farm structures and management practices, with more than 90% of the farms participating.  Assessment of the impacts of the SLWAP on water quality is challenging; aggressive development of whole farm plans and implementation of best management practices (BMPs) precludes a classic pre- vs. post-implementation analysis.  We used a combination of stream water monitoring and watershed modeling to assess program impacts on water quality. Stream flow and concentrations of sediment and nutrients were monitored over a 3-year period for 6 major subwatersheds.   The AnnAGNPS model was used to predict sediment and TN loads at the watershed outlets. The model allows one to distribute loads for individual cells comprising each subwatershed.  This procedure, referred to as source accounting, allowed us to track pollutant loads to individual cells with a specific land use.  Discriminant analysis, applied to individual cell data output from AnnAGNPS, failed to differentiate forest from agricultural land use on the basis of sediment and total N loads.  We interpret these results as an indication of BMP effectiveness in maintenance of water quality in agriculturally managed lands in the Skaneateles Lake Watershed.