Abstract Sex estimation is the first step for biological profile reconstruction of an unknown skeleton (archaeological or contemporary) and consequently for positive identification of skeletal remains recovered from forensic settings. Several tools have been developed using different osseous structures. With the intention to provide an objective method comparison, we reported the analysis of three different methods (visual, metric and geometric morphometrics) for sex assessment of the greater sciatic notch. One hundred and thirty pelvic bones (45.4% females and 54.6% males) from the National Autonomous University of Mexico Skeletal Collection pertaining to the contemporary Mexican population were analyzed. We used the ROC-analysis to test between desired false positive thresholds (1-specificity) and expected true positive rates (sensitivity) in order to predict the best approach to sex assessment. The comparison of the area under the ROC-curves shows significant differences among visual and metric methods. At the same time, the analysis suggested that higher morphological variation among the sexes is independent of the methodological approach. The results indicate that the metric (angle), with a high percent of indeterminate cases (34.6%), and visual, with 26.2% of the cases allocated as intermediate cases, were poorly accurate; we cannot recommend these techniques for sexing an unknown specimen. On the other hand, the geometric morphometrics approach improves sex estimation in 82.3% of correctly classified individuals with more than 95% of posterior probability. In addition to the method comparison, the major sexual variation of the greater sciatic notch was determined to be located on its posterior border.