Lake Michigan surface waters impacted by fecal pollution were assessed to determine the occurrence of genetic markers for Bacteroides and Escherichia coli. Initial experiments with sewage treatment plant influent demonstrated that total Bacteroides spp. could be detected by PCR in a 25- to 125-fold-higher dilution series than E. coli and human-specific Bacteroides spp., which were both found in similar dilution ranges. The limit of detection for the human-specific genetic marker ranged from 0.2 CFU/100 ml to 82 CFU/100 ml culturable E. coli for four wastewater treatment plants in urban and rural areas. The spatial and temporal distributions of these markers were assessed following major rain events that introduced urban storm water, agricultural runoff, and sewage overflows into Lake Michigan. Bacteroides spp. were detected in all of these samples by PCR, including those with <1 CFU/100 ml E. coli. Human-specific Bacteroides spp. were detected as far as 2 km into Lake Michigan during sewage overflow events, with variable detection 1 to 9 days postoverflow, whereas the cow-specific Bacteroides spp. were detected in only highly contaminated samples near the river outflow. Lake Michigan beaches were also assessed throughout the summer season for the same markers. Bacteroides spp. were detected in all beach samples, including 28 of the 74 samples that did not exceed 235 CFU/100 ml of E. coli. Human-specific Bacteroides spp. were detected at three of the seven beaches; one of the sites demonstrating positive results was sampled during a reported sewage overflow, but E. coli levels were below 235 CFU/100 ml. This study demonstrates the usefulness of non-culture-based microbial-source tracking approaches and the prevalence of these genetic markers in the Great Lakes, including freshwater coastal beaches.