We examined the feasibility of using repeated sampling with binary detection of genetic markers as a means to categorize fecal pollution amongst three host sources at ten sample locations in the Colville River watershed (Washington State, U.S.A). Water samples from each site were collected eight times between June and August 2005. Total enterococci were enumerated after membrane filtration and resulting colonies were recovered for PCR testing with a panel of 16 genetic markers. The geometric mean for enterococci ranged from 38 to 945 CFU. Six of the markers were positive for all samples during the first three sample visits and four of the human markers were rarely encountered. Six markers identified four sites as impacted by human waste, four sites as impacted by cattle waste, and three sites as impacted by cervid waste. There are no gold standards by which to assess the diagnostic sensitivity and specificity of BST markers in the field, but results from our sampling scheme were consistent with proximal land-use practices and permitted relative rankings between sites in a cost-effective manner.