We compared survey techniques for estimating relative and absolute abundances of swift foxes (Vulpes velox) in New Mexico. For relative abundance surveys, the most efficient technique is collection of scats followed by verification of species depositing scats with DNA analysis. By collecting scats, the proportions of individual locations where swift foxes were detected were 61.9% and 67.7% during surveys in 2000 and 2001, which were greater than the proportions using scent stations (31.4%, 47.1%) or trapping (11.5%, 8.4%). By collecting scats, we detected swift foxes in 100% of the fox home ranges within the study area. If scent-station surveys are used instead, scent-station transects consisting of stations spaced at 1.6 km (1.0 mile) intervals and operated for three nights are the most practical. Searching for tracks, spotlighting and calling are much less efficient techniques. For absolute abundance surveys, trapping and resighting with cameras at bait stations was more accurate than counting unique microsatellite DNA genotypes from collected scats. Using trapping/resighting, we estimated the 95% confidence intervals for the swift fox population within the study area to be 17.8–30.0, 11.9–25.3 and 15.2–17.3 in the periods November 1999–January 2000, February 2000 and January–March 2001, respectively. We counted 63 and 27 unique genotypes in early 2000 and 2001, respectively. The numbers of unique genotypes, which were much greater than population estimates obtained from trapping and resighting, were overestimated because of the presence of transient swift foxes and poor quality DNA from scats leading to allelic drop-out and/or false alleles.