Microbiological source tracking (MST) methods are increasingly being used to identify fecal contamination sources in surface waters, but these methods have been subjected to limited comparative testing. In this study, 22 researchers employing 12 different methods were provided sets of identically prepared blind water samples. Each sample contained one to three of five possible fecal sources (human, dog, cattle, seagull or sewage). Researchers were also provided with portions of the fecal material used to inoculate the blind water samples for use as library material. No MST method that was tested predicted the source material in the blind samples perfectly. Host-specific PCR performed best at differentiating between human and non-human sources, but primers are not yet available for differentiating between all of the non-human sources. Virus and F+ coliphage methods reliably identified sewage, but were unable to identify fecal contamination from individual humans. Library-based isolate methods correctly identified the dominant source in most samples, but also had frequent false positives in which fecal sources not in the samples were incorrectly identified as being present. Among the library-based methods, genotypic methods generally performed better than phenotypic methods.