Background: High morphological diversity can occur in closely related animals when selection favors morphologies that are subject to intrinsic biological constraints. A good example is subterranean rodents of the genus Thomomys, one of the most taxonomically and morphologically diverse mammalian genera. Highly procumbent, tooth-digging rodent skull shapes are often geometric consequences of increased body size. Indeed, larger-bodied Thomomys species tend to inhabit harder soils. We used geometric morphometric analyses to investigate the interplay between soil hardness (the main extrinsic selection pressure on fossorial mammals) and allometry (i.e. shape change due to size change; generally considered the main intrinsic factor) on crania and humeri in this fast-evolving mammalian clade. Results: Larger Thomomys species/subspecies tend to have more procumbent cranial shapes with some exceptions, including a small-bodied species inhabiting hard soils. Counter to earlier suggestions, cranial shape within Thomomys does not follow a genus-wide allometric pattern as even regional subpopulations differ in allometric slopes. In contrast, humeral shape varies less with body size and with soil hardness. Soft-soil taxa have larger humeral muscle attachment sites but retain an orthodont (non-procumbent) cranial morphology. In intermediate soils, two pairs of sister taxa diverge through differential modifications on either the humerus or the cranium. In the hardest soils, both humeral and cranial morphology are derived through large muscle attachment sites and a high degree of procumbency. Conclusions: Our results show that conflict between morphological function and intrinsic allometric patterning can quickly and differentially alter the rodent skeleton, especially the skull. In addition, we found a new case of convergent evolution of incisor procumbency among large-, medium-, and small-sized species inhabiting hard soils. This occurs through different combinations of allometric and non-allometric changes, contributing to shape diversity within the genus. The strong influence of allometry on cranial shape appears to confirm suggestions that developmental change underlies mammalian cranial shape divergences, but this requires confirmation from ontogenetic studies. Our findings illustrate how a variety of intrinsic processes, resulting in species-level convergence, could sustain a genus-level range across a variety of extrinsic environments. This might represent a mechanism for observations of genus-level niche conservation despite species extinctions in mammals.