Bat fatality monitoring at wind turbines depends upon reliable identification of carcasses. Using reference mitochondrial cytochrome c oxidase I gene sequences mined from GENBANK and new sequences from collected samples, we constructed maximum likelihood trees including all 47 bat species found in the USA and tested the use of this locus for DNA barcoding these bat species. In this study, 80 % of species examined had distinct barcodes, including species currently listed as threatened or endangered. Nine of 17 Myotis bat species examined did not form distinct clades and had very low inter-specific genetic distances (1.4 %), thereby making this barcoding technique unreliable for some members of this genus. We then applied this technique to DNA samples from 892 bats salvaged from wind farms across four states. Using DNA barcoding, we were able to identify 14 carcasses to species that could not be identified in the field due to extensive decomposition and scavenging, and determined that another 18 carcasses had been misidentified in the field. Furthermore, we found field misidentifications increased with time until discovery. We conclude that DNA barcoding can improve the identification of salvaged bat carcasses especially when rare and uncommon species are encountered. This technique has other practical applications, such as identifying remains from hibernacula (potentially including carcasses of unknown bats with white-nose syndrome) or identifying species from fecal samples at roost sites or other locales.