Estimating body mass/size/weight remains a crucial precursor to the evaluation of relative brain size and to achieving an understanding of brain evolution in fossil species. Despite the obvious close association between the metrics of postcranial elements and body mass a number of factors combine to reduce their utility. This study examines the feasibility of cranial variables for predicting body mass. The use of traditional regression procedures, independent contrasts analysis, and variance partitioning all support the hypothesis that cranial variables are correlated with body mass even when taking phylogeny into account, with r values typically ranging between 0.52 and 0.98. Body mass estimates derived for fossil hominins using cranial variables are similar to those obtained from previous studies using either cranial or postcranial elements. In particular, upper facial breadth and orbital height display strong predictive capability. Average body masses derived from Least Squares Regression (LSR) equations were used to calculate estimates of body mass for three hominin species. This resulted in estimates of between 30 kg and 47 kg for Australopithecus africanus, 48 kg and 52 kg for Paranthropus robustus, and 75 kg for Homo neanderthalensis. It is proposed that regression equations derived for the order primates are used to estimate body mass for archaic hominins, while hominoid based equations are most suited for Homo. Am J Phys Anthropol, 2007. © 2007 Wiley-Liss, Inc.