The number and delimitation of species in the Phyciodes tharos species complex has puzzled lepidopterists for years. Previous analysis of mtDNA sequence data has suggested that P. cocyta is more closely related to P. cocyta than to P. tharos, in contrast to inferences from morphology and ecology. We sequenced the mitochondrial gene cytochrome oxidase I for 40 individuals of Phyciodes tharos, P. cocyta, and P. cocyta from Michigan and Ohio, a region at the boundaries of the geographic ranges of these species. Network and cladistic analyses reveal shared mtDNA lineages, indicating that limited hybridization occurs in this region between P. cocyta and the other two taxa but not between P. tharos and P. cocyta. Our evidence also supports the traditional phylogenetic assessment of P. tharos and P. cocyta as the two most closely related species in this species group. Data from nuclear genes are needed to more fully resolve this intriguing group of butterflies.