We used a library of DNA fingerprints, created using the rep-PCR and HFERP techniques, in an attempt to define sources of fecal bacterial pollution, E. coli, in three Minnesota watersheds, Minneopa Creek (Blue Earth County), High Island Creek (Sibley County), and Vermillion River (Dakota County). Sampling from 10 sites per watershed took place in 2001 and 2002. Approximately 25 E. coli isolates were obtained from each site per sampling date. About 1,776, 1,651, and 1,762 E. coli were DNA fingerprinted from the Vermillion River, High Island Creek, Minneopa Creek Watersheds, respectively. The most reliable results from data came from bootstrap analyses of fecal bacteria segregated into Human vs. Non-human categories, or into groupings consisting of Humans, Pets (dogs and cats), Waterfowl (geese, ducks), Wildlife (deer), and Domesticated animals (chickens, cows, goats, horses, pigs, sheep, turkeys). Analysis of the Vermillion River showed that 93 and 6.1 % of the isolates identified were of Non-Human and Human origin, respectively. The greatest potential contributors to fecal pollution in this watershed were domesticated animals (23 %), pets (45%), and deer (19%). Similar results were found with the Minneopa Creek isolates, where 90 and 10% of the isolates were from non-human and human origin, respectively. Of these 23% were from Domesticated animals, 36% from Pets, and 21% from deer. In contrast, while 84 and 16% of High Island Creek isolates were Non-Human and Human sources, respectively, the majority came from domesticated animals (42%, mostly from cows), with the remainder contributed by geese, 14%, and humans 16%. It should be noted however, that our research showed that much larger database of DNA fingerprints is needed for more accurate assignments to the animal level. A reliable bacterial source tracking method would aid watershed managers tremendously, giving them another tool to efficiently direct efforts to clean watersheds of bacterial pollutants.