Surface-water impairment by fecal coliform bacteria is a water-quality issue of national scope and importance. In Virginia, more than 175 stream segments are on the Commonwealth's 1998 303(d) list of impaired waters because of elevated concentrations of fecal coliform bacteria. These fecal coliform-impaired stream segments require the development of total maximum daily load (TMDL) and associated implementation plans, but accurate information on the sources contributing these bacteria usually is lacking. The development of defendable fecal coliform TMDLs and management plans can benefit from reliable information on the bacteria sources that are responsible for the impairment. Bacterial source tracking (BST) recently has emerged as a powerful tool for identifying the sources of fecal coliform bacteria that impair surface waters. In a demonstration of BST technology, three watersheds on Virginia's 1998 303(d) list with diverse land-use practices (and potentially diverse bacteria sources) were studied. Accotink Creek is dominated by urban land uses, Christians Creek by agricultural land uses, and Blacks Run is affected by both urban and agricultural land uses. During the 20-month field study (March 1999?October 2000), water samples were collected from each stream during a range of flow conditions and seasons. For each sample, specific conductance, dissolved oxygen concentration, pH, turbidity, flow, and water temperature were measured. Fecal coliform concentrations of each water sample were determined using the membrane filtration technique. Next, Escherichia coli (E. coli) were isolated from the fecal coliform bacteria and their sources were identified using ribotyping (a method of 'genetic fingerprinting'). Study results provide enhanced understanding of the concentrations and sources of fecal coliform bacteria in these three watersheds. Continuum sampling (sampling along the length of the streams) indicated that elevated concentrations of fecal coliform bacteria (maximum observed concentration of 290,000 colonies/100 milliliters (col/100mL) could occur along the entire length of each stream, and that the samples collected at the downstream monitoring station of each stream were generally representative of the entire upstream reach. Seasonal patterns were observed in the base-flow fecal coliform concentrations of all streams; concentrations were typically highest in the summer and lowest in the winter. Fecal coliform concentrations were lowest during periods of base flow (typically 200?2,000 col/100mL) and increased by 3?4 orders of magnitude during storm events (as high as 700,000 col/100mL). Multiple linear regression models were developed to predict fecal coliform concentrations as a function of streamflow and other water-quality parameters. The source tracking technique provided identification of bacteria contributions from diverse sources that included (but were not limited to) humans, cattle, poultry, horses, dogs, cats, geese, ducks, raccoons, and deer. Seasonal patterns were observed in the contributions of cattle and poultry sources. There were relations between the identified sources of fecal coliform bacteria and the land-use practices within each watershed. There were only minor differences in the distribution of bacteria sources between low-flow periods and high-flow periods. A coupled approach that utilized both a large available source library and a smaller, location-specific source library provided the most success in identifying the unknown E. coli isolates. BST data should provide valuable support and guidance for producing more defendable and scientifically rigorous watershed models. Incorporation of these bacteria-source data into watershed management strategies also should result in the selection of more efficient source-reduction scenarios for improving water quality.