Recent work has revealed surprisingly deep mitochondrial genetic divergence in the lizard Anolis cristatellus among samples obtained from the small Caribbean island of Vieques. Here we sought to determine whether this had resulted from natural or anthropogenic causes, and (if the former) whether divergence occurred in a biogeographical context of allopatry followed by secondary contact, or via isolation-by-distance across the species’ historical range. We first estimated a mitochondrial gene tree for 379 samples and then genotyped 3407 single nucleotide polymorphic sites from 48 individuals using a modified genotyping-by-sequencing approach. We found that A. cristatellus samples from Vieques belong to two highly divergent mitochondrial subclades, but the geographical distribution of these haplogroups indicates that this pattern is probably natural in origin. Analysis of our single nucleotide polymorphic dataset revealed differentiation that is consistent with isolation-by-distance between the western and eastern ends of Vieques, suggesting that the overall pattern of divergence probably reflects primary intergradation with a mitochondrial break on the historical Puerto Rico Bank palaeo-island that happened to coincide with the present-day location of Vieques. Our findings help to underline the growing consensus that results from a single genetic marker can prove highly misleading in studies of historical population genetic structure.