Carnivores typically require large areas of habitat, exist at low natural densities, and exhibit elusive behavior - characteristics that render them difficult to study. Noninvasive survey methods increasingly provide means to collect extensive data on carnivore occupancy, distribution, and abundance. During the summers of 2003-2004, we compared the abilities of scat detection dogs, remote cameras, and hair snares to detect black bears (Ursus americanus), fishers (Martes pennanti), and bobcats (Lynx rufus) at 168 sites throughout Vermont. All 3 methods detected black bears; neither fishers nor bobcats were detected by hair snares. Scat detection dogs yielded the highest raw detection rate and probability of detection (given presence) for each of the target species, as well as the greatest number of unique detections (i.e., occasions when only one method detected the target species). We estimated that the mean probability of detecting the target species during a single visit to a site with a detection dog was 0.87 for black bears, 0.84 for fishers, and 0.27 for bobcats. Although the cost of surveying with detection dogs was higher than that of remote cameras or hair snares, the efficiency of this method rendered it the most cost-effective survey method.