Water quality in surface waters, ponds and lakes is frequently degraded by fecal contamination from human and animal sources, imposing negative implications for recreational uses and public safety. For this reasons it is critical to accurately identify the source of fecal contamination in water bodies in order to take corrective actions for fecal pollution control. Bacteroides-specific molecular markers have been widely used to differentiate human from other sources of fecal bacteria in waters. Our previous work indicated that many assays currently used to detect human-specific Bacteroides produce false positive results in the presence of fish (tilapia, catfish, trout, salmon) fecal DNA. Bacteroides 16S rRNA gene sequences obtained from fish and other selected animals aided us to achieve our objective for developing a new microbial source tracking strategy targeting a human-specific Bacteroides molecular marker. We have successfully cultured, speciated and identified Bacteroides from fish and human fecal samples. Furthermore, Bacteroides 16S rRNA gene sequences from fish and selected animals were aligned against human B. volgatus, B. fragilis and B. dorei to identify variable regions within the 16S rRNA. The data was used to develop multi-segmented variable regions for differentiating human versus animal sources. Additionally, the 16S rRNA gene sequences from variable regions were connected together (truncated sequences) and were used to construct phylogenetic tree. The proposed strategy was field tested by collecting water samples from central AZ sources and from ponds. PCR using HF134 and HF183 primers were performed and sequences for positive reactions were then aligned against human Bacteroides to identify the source of contamination. Forty six Bacteroides 16S rRNA gene sequences from fish were deposited at the NCBI database. Four variable regions within the 16S rRNA gene sequence were identified as regions 1 (71-101), 2 (142-271), 3 (451-511) and 4 (581-701). The phylogenetic tree constructed from truncated sequences shows distinct clades for each animal. By using HF134 and HF183 primers, the source of fecal contamination was presumptively identified. For the samples with positive signals (8/13), the source of fecal contamination can be from human. For confirming the data, the sequences from PCR products must first be aligned against the four variable regions and then compare with the truncated phylogenetic tree. It is expected for human sequences to be grouped within the human clade. We have identified a sufficient variability matrix based on the four regions and a clear grouping within the truncated phylogenetic tree which is sufficient to differentiate Bacteroides isolates. The proposed strategy offers a new method for microbial source tracking and provides step-wise methodology which is essential for identifying sources of fecal pollution.