I evaluated the use of non-invasively collected hair samples from white-tailed deer (Odocoileus virginianus) in three contexts. First I assessed the effects of sampling interval and barb location on the probability of sample cross-contamination of hair snares. Six hair snares were installed on Presque Isle Park, Marquette Michigan from 11 May 2008 to 3 July 2008 and hair was collected from each daily. Probability cross-contamination increased from 12% to 28% during sampling intervals of two and seven days, respectively, but was unaffected by barb location. Second I assessed the benefits and costs of using non-invasive techniques for population estimation. I estimated abundance of the Presque Isle deer herd from genotype data derived from barbed-wire snared hair, and compared these estimates to drive counts performed during this study. Genotype-based estimates were greater than the drive counts, probably due to the genetic “capture” of animals that frequently move on and off the peninsula. This represents the first successful use of non-invasive genetic sampling for population estimation of ungulates. Third, I compared the amount of genetic differentiation and migration between two island-mainland systems in Lake Superior. Results differed from a standard biogeographical prediction that populations on larger islands closer to their mainland source population should have higher indices of genetic connectivity. The results of this research should be used to inform future studies that use non-invasive genetic sampling for ungulate population research.