Craniofacial remains (the most abundant identifiable remains in the fossil record) potentially offer important information about body size dimorphism in extinct species. This study evaluates the scaling relationships between body mass dimorphism and different measures of craniofacial dimorphism, evaluating taxonomic differences in the magnitude and scaling of craniofacial dimorphism across higher taxonomic groups. Data on 40 dimensions from 129 primate species and subspecies demonstrate that few dimensions change proportionally with body mass dimorphism. Primates show general patterns of greater facial vs. neurocranial and orbital dimorphism, and greater dimorphism in lengths as opposed to breadths. Within any species, though, different craniofacial dimensions can yield very different reconstructions of size dimorphism. There are significant taxonomic differences in the relationships between size and craniofacial dimorphism among primate groups that can have a significant impact on reconstructions of body mass dimorphism. Hominoids tend to show lower degrees of facial dimorphism proportional to size dimorphism than other primates. This in turn implies that strong craniofacial dimorphism in Australopithecus africanus could imply very strong body size dimorphism, conflicting with the relatively modest size dimorphism inferred from postcrania. Different methods of estimating the magnitude of size dimorphism from craniofacial measurements yield similar results, and yield comparatively low percent prediction errors for a number of dimensions. However, confidence intervals for most estimates are so large as to render most estimates highly tentative. Am J Phys Anthropol 120:38–60, 2003. © 2003 Wiley-Liss, Inc.