Abstract: The most accurate and precise methods for the assessment of age and stature often require knowledge of sex. Thus, being able to correctly identify sex from skeletal remains is critical in the forensic context. The presence of the os coxae or skull can never be guaranteed, making the development of reliable methods of sex estimation using other skeletal elements necessary. Using a 724 individual calibration sample from the Hamann-Todd collection, this study identifies sexual dimorphism in the human scapula, and presents a new five-variable discriminant function for sex estimation. The overall accuracy of this method proved to be 95.7% on the cross-validated calibration sample, 92.5% on an 80 individual test sample from the Hamann-Todd collection, and 84.4% on a 32 individual test sample from the skeletal collection of the Wichita State University Biological Anthropology Laboratory. Additionally, a slightly less accurate two-variable model was developed and has cross-validated accuracy of 91.3%.