SUMMARY: Previous studies have yielded very different results in sex estimation based on measurements of the lateral angle (LA) of the temporal bone. The purpose of this study was to, first, investigate if the bad results obtained by the LA method could be due to the methodology and then, second, to examine sexual dimorphism in the relationship between the lateral angle and cranial base shape. The lateral angle method was tested using a forensic sample of 102 CT scans of the head with known sex. We measured the angle using two methods: measurements directly on the CT slide, the method usually applied, and by use of a new method, using a "virtual cast". The cranial base was quantified by placing 12 landmarks in the posterior fossa. Procrustes analysis, principal component analysis, discriminant analysis and cross-validation test were performed. The "cast method" was found to be less accurate than the direct measurements. The mean angle was greater in females (48.2° ± 7.2°) than in males (45.38° ±8.06°) but the difference was not significant (t-test, p = 0.063). A statistically significant difference in cranial base shape existed between the two sexes, but the results also demonstrated a major overlap between the female and male shapes. In the light of the observed results, we confirm the lack of accuracy reported when using the lateral angle method to determine sex and we hypothesise that it may be explained by a lack of sexual dimorphism of the cranial base in terms of shape.