As part of the University of Michigan Dioxin Exposure Study, soil, household dust, and serum samples were collected from more than 750 households in five populations around the city of Midland and in Jackson and Calhoun Counties, Michigan, USA. Polytopic vector analysis, a type of receptor model, was applied to better understand the potential sources of polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans found in these samples and to quantify the contributions of the sources in each matrix across populations. The results indicated that source signatures found in soil are similar to those found in dust, reflecting various combustion profiles, pentachlorophenol, and graphite electrode sludge. The profiles associated with contamination in the Tittabawassee River, likely related to historical discharges from the Dow Chemical Company facility in Midland, exhibited the largest differences among the regional populations sampled. Differences in serum source contributions among the study populations were consistent with some of the regional differences observed in soil samples. However, the age trends of these differences suggested that they are related to past exposures, rather than ongoing sources.