Cannell [Journal of Archaeological Science 29:335–339, 2002] argued that sex-based differences among humans in terms of the mass of chosen throwing stones could be used to infer body mass and patterns of sexual dimorphism in early hominids from Olduvai and Koobi Fora by examining the mass distributions of unaltered stone tools at those sites. We examined this hypothesis in tufted capuchin monkeys using a comparative approach, by investigating the relationships among body mass, sex, stone weight preference, and accuracy in a throwing task. The subject sample consisted of nine monkeys trained to perform an aimed-throwing task in which a food reward could be obtained by throwing a stone into a bucket. We found that 1) the subjects showed a strong mean stone mass preference; 2) the females chose heavier stones than the males, in terms of absolute mean selected stone mass and selected stone mass relative to body mass; 3) subjects threw more accurately when they used stones of preferred mass vs. stones of nonpreferred mass; and 4) overall, the males were more accurate in the throwing task than the females. We conclude that capuchins are highly selective when choosing throwing stones, and that this confers an advantage for throwing accuracy. Our results indicate that the sexually dimorphic pattern in stone mass preference observed among humans does not generalize to Cebus apella. We suggest that researchers examining this pattern in humans in an attempt to explain early hominid patterns of dimorphism and behavior should take into account not only stone weight preference, but also its adaptive advantage. Am. J. Primatol. 61:159–172, 2003. © 2003 Wiley-Liss, Inc.