The present study examines the sexing potential of the minimum supero-inferior femoral neck diameter in Caucasians and African-Americans who lived at the turn of the century. A Student's t-test and an ANOVA indicate that population differences in neck morphology exist, albeit the strength of the test is fairly weak (P ฯญ 0.015). Predictive models were developed using a linear discriminant function analysis for the African-American sample, the Caucasian sample, and the combined African-American and Caucasian (AAC) sample. Jackknifed classification matrices produced classification success rates ranging from 87 to 92%. Each of the three discriminant functions were evaluated using an independent, random holdout sample. Although a smaller holdout sample usually better approximates the true error involved in an application, this was clearly not the case in this study. For African-Americans, 28 of 28 individuals were correctly classified, for Caucasians 24 of 25, and for the combined AAC sample 53 of 53 individuals were sexed correctly. It is more likely that the true accuracy of the model for the population approximates 90%. This accuracy combined with the high rate of preservation of the femoral neck makes this measurement useful in extremely fragmentary samples.