Although genetic and analytical methods for estimating wildlife abundance have improved rapidly over the last decade, effective methods for collecting hair samples from terrestrial carnivores in a mark–recapture framework have lagged. Hair samples are generally collected using methods that permit sampling of multiple individuals during a single sampling period that can cause genotyping errors due to cross-contamination. We evaluated a modified body snare as a single-sample method to obtain bobcat hair samples suitable for individual identification using DNA analyses to estimate population size. We used a systematic grid (2.5 × 2.5 km) overlaid on a 278.5 km2 study area in Michigan's Upper Peninsula to distribute sampling effort. In each of 44 grid cells, we placed 2–6 snares at established sampling stations and collected hair samples weekly for 8 weeks during January–March 2010. We collected 230 hair samples overall, with 91% of sampling stations obtaining at least 1 hair sample. Fifty-seven percent of samples had sufficient DNA for species identification, which included bobcat (Lynx rufus, n = 17); raccoon (Procyon lotor, n = 62); coyote, dog, or wolf (Canis spp., n = 29); fox (Vulpes vulpes or Urocyon cinereoargenteus, n = 4); and fisher (Martes pennanti, n = 1). We identified 8 individual bobcats and using Huggins closed capture population models with a one-half mean maximum distance moved buffer, estimated 10 individuals within the trapping area (95% confidence interval = 8–28) with a density of 3.0 bobcats/100 km2. Our method provides an effective, single-sample technique for detecting bobcats and estimating abundance. © 2012 The Wildlife Society.