The authors constructed both univariate and multivariate discriminant models using data from the Forensic Anthropology Data Bank. Discriminating effectiveness was assessed by cross-validated classification, and in the case of multivariate models, Mahalanobis D(2). The results clearly indicate that most postcranial elements outperform the skull in estimating sex. It is possible to correctly sex 88-90 percent of individuals with joint size, up to 94 percent with multivariate models of the postcranial bones. The best models for the cranium do not exceed 90 percent. The authors conclude that postcranial elements are to be preferred to the cranium for estimating sex when the pelvis is unavailable. (Published Abstract)