The increasing use of mitochondrial DNA (mtDNA) to explore and test species limits among morphologically similar species is potentially compromised by phenomena poorly reflective of organismal history and speciation, including (but not limited to) stochastic lineage sorting and gene flow. In situations where molecular data are only available from a single gene or linkage partition (e.g. mtDNA), corroboration of suspected species boundaries should be sought from independent lines of evidence, such as morphology. Recent attempts to delimit species using mtDNA and morphology have either implicitly or explicitly ignored the possibility that distinct species can occur in direct sympatry throughout much of their range, presumably because such situations are believed to be rare. We examined phylogenetic relationships within the long-tailed shrew tenrecs (Mammalia: Tenrecidae; Microgale spp.) from Madagascar. Current taxonomy recognizes two broadly sympatric species, though as many as six have been described. Given that alpha taxonomy within shrew tenrecs has been controversial, and that patterns of morphological variation can be especially difficult to assess for this group, some authors have suggested that additional cryptic species may exist. To examine this possibility, we conducted a phylogenetic study using the mitochondrial NADH dehydrogenase subunit 2 gene and a morphometric analysis of 29 craniodental, postcranial, and external measurements from a broad geographical sample of long-tailed shrew tenrecs. The two data sets were nearly perfectly congruent in identifying four groups that can be classified as species, thereby doubling the currently recognized number of species. We present previously unrecognized distributional evidence consistent with our conclusions and provide an empirical example of how a revised understanding of species limits alters inferences of geographic variation and species coexistence, particularly with respect to fine-scale habitat partitioning. The results of this study suggest that certain species pairs, previously assumed to be single species occupying broad elevational ranges, are actually reproductively isolated units that are partitioning their environment along elevational lines. © 2004 The Linnean Society of London, Biological Journal of the Linnean Society, 2004, 83, 1–22.