In a recent paper, Kambhampati, Clark & Brock (Biological Journal of the Linnean Society 2002; 75: 163–172) suggested that members of the wood‐feeding cockroach genus Cryptocercus are differentially adapted to two features of the environment: host log species and elevation. A re‐examination of the evidence, however, fails to support their hypothesis. First, their analysis of host preferences was based on a general description of forest type, rather than on the level at which host choice occurs: the dead tree on the forest floor. Cryptocercus in both East Asia and in the eastern United States have been collected from a range of evergreen and deciduous logs. Although C. clevelandi in the western United States is associated primarily with conifers, no evidence of host adaptation exists. Second, there is no support for the described evolutionary trend toward low‐altitude habitats among karyotype groups of C. punctulatus in the eastern United States. The findings of Kambhampati et al. are based on inadequate sampling; they did not include the highest and most topographically complex regions of the Southern Appalachian Mountains, nor the lower elevation border regions to the north and east of the range. An analysis of elevational data from 71 collection sites revealed no significant difference in the range of altitudes at which karyotype groups of C. punctulatus are found. In contrast to the suggestions of these authors, a lack of specialization with regard to altitude and host logs was probably the factor that allowed all taxa in the genus to thrive during climatically driven shifts in the geographical location and plant species composition of northern hemisphere mesic forests. © 2003 The Linnean Society of London, Biological Journal of the Linnean Society, 2003, 80, 223–233.