Craniodental robusticity in Paranthropus has led many researchers to posit that all the species in this genus share a common adaptation to a diet of hard foods. Recent research on craniodental morphology, microwear, biomechanics, and isotopes, by contrast, has suggested that substantial variation exists within the genus Paranthropus, both in terms of ecological niches occupied by the three recognized species within the genus and the amount of consumed hard and compliant foods. Rather than pointing to a common adaptive suite, these studies suggest that the species were adaptively distinct from each other. However, current approaches to understanding craniodental morphology do not present a clear picture of how these speciesspecific adaptations differ. It is also not clear whether all aspects of morphology that have been attributed to adaptation are indeed adaptive, rather than the products of non-adaptive processes. This study examines variation across the three known Paranthropus taxa (P. aethiopicus, P. boisei and P. robustus; N=39) using an approach that tests for adaptive morphology against a null hypothesis of random change (i.e. drift). Extant species (Homo sapiens (N=150), Gorilla gorilla (N=150), Pan troglodytes (N=143) act as analogues for Paranthropus variance/covariance (V/CV). Results reveal a high magnitude of variation within and between species across mandibular and cranial regions, especially when including the P. robustus individuals DNH 7 & 8 from Drimolen. Drift cannot be rejected for the bulk of comparisons. Neutrality tests detect adaptive divergence between P. robustus and the other two species, but not between P. aethiopicus and P. boisei. Reconstructed selection vectors indicate that both positive and negative directional selection have driven diversification in mandibular and tooth dimensions and in the cranium, resulting in variable morphological responses including considerable evidence for correlated selection.