The evolution of wing pattern diversity in butterflies has emerged as a model system for understanding the origins and maintenance of adaptive phenotypic novelty. Admiral butterflies (genus Limenitis) are an attractive system for studying wing pattern diversity because mimicry is common among the North American species and hybrid zones occur wherever mimetic and non-mimetic wing pattern races meet. However, the utility of this system has been limited because the evolutionary relationships among these butterflies remain unclear. Here I present a robust species-level phylogeny of Limenitis based on 1911 bp of two mitochondrial genes (COI and COII) and 904 bp of EF1-alpha for all five of the Nearctic species/wing pattern races, the majority of the Palearctic species, and three outgroup genera; Athyma, Moduza (Limenitidini), and Neptis (Limenitidinae: Neptini). Maximum-likelihood and Bayesian analyses indicate that the North American species are a well-supported, monophyletic lineage that is most closely related to the widespread, Palearctic, Poplar admiral (L. populi). Within North America, the Viceroy (L. archippus) is the basal lineage while the relationships among the remaining species are not well resolved. A combined maximum-likelihood analysis, however, indicates that the two western North America species (L. lorquini and L. weidemeyerii) are sister taxa and closely related to the wing pattern subspecies of the polytypic Limenitis arthemis species complex. These results are consistent with (1) an ancestral host-shift to Salicaceae by the common ancestor of the Poplar admiral and the Nearctic admiral lineage, (2) a single colonization of the Nearctic, and (3) a subsequent radiation of the North American forms leading to at least three independent origins of mimicry.