Discriminant functions were developed using long-bone robusticity measurements of 82 individuals from the protohistoric Maya site of Tipu, Belize. All individuals were sexed using nonmetric morphological indicators, particularly those of the pelvis. These functions are designed to provide a means of determining the sex of fragmentary prehistoric Maya skeletons. The equations ranged in accuracy from 77.5% to 100%. The reliability of these equations was tested using a jackknife method on the Tipu sample and by applying the equations to small samples of prehistoric skeletons from the sites of Seibal, KOB Swamp, Laguna de On, and Chau Hiix. The vast majority of the equations applied to the test cases succeeded in correctly estimating the sex based on pelvic and cranial features. A more reliable technique for sex determination of poorly preserved skeletal remains will allow a whole new range of archaeological and bioanthropological hypotheses concerning sex and gender among the ancient Maya to be investigated and considered.