Species form the fundamental units of analysis in many areas of biology and, therefore, rigorous delimitation of this unit is important to a broad array of researchers. Recently, many new empirical methods have been proposed to delimit species in nature, and a large literature exists on the theoretical merit and superiority of each method. However, few empirical studies actually compare the results of these methods applied in the same study system. We used a large allozyme and chromosome dataset to apply a number of genetic-distance, character-based, and tree-based methods to a well-studied, data-rich system: the Sceloporus grammicus lizard complex of central Mexico. We hypothesized species boundaries under a general lineage or evolutionary species conceptual framework in an a priori fashion using mapped restriction-site data (mitochondrial DNA and nuclear rDNA), allozymes, and morphology. We then compared the ability of different methods to recover the “hypothesized evolutionary species” (HES). Highton's genetic-distance method and a tree-based method consistently recovered all four HES, although sometimes with weak support. With two exceptions, other methods recovered the same HES, but additional groups were weakly delimited and nested within the HES. Given the apparent recent divergence of some of the chromosome races and distinct populations in this complex, these are encouraging results. We emphasize the value of specifying testable criteria as clearly as possible and testing these with methods that make use of different properties of a single dataset.