Loth & Henneberg (Am J Phys Anthrop 99 (1996) 473) described a single morphological indicator of sexual dimorphism, namely the presence or absence of flexure on the posterior border of the mandible, with a predictive accuracy of 90.6-99.0%. In the other studies, which have criticized mandibular ramus flexure as sex indicator in adult and fossils specimens by the same method, the accuracy of sexing was found between 59.0% and 80.4% which is well below the reported 90.6-99.0% (Am J Phys Anthrop 107 (1998) 363; Am J Phys Anthrop 111 (2000) 573; Am J Phys Anthrop 111 (2000) 429; Homo 53 (2002) 97). It was indicated that the method sexed males more reliably than females and consistency was low. In the current blind test, 120 mandibles from forensic cases were examined for the presence or absence of mandibular ramus flexure. Virtually identical results were obtained when a second observer examined the same sample of mandibles. The results were then compared to records of the same cases resulting in 85.8% accuracy for the total sample. A marked difference was observed between sexes where the ramus shape was diagnostic for 92.6% of the males but for only 60.0% of the females (p < 0.001). When mandibles with excessive tooth loss (more than two molars missing) (n = 35) were removed from the sample (in normative sample), the overall predictive value increased to 90.6% (95.6% accuracy in males and 70.6% in females). Our results confirm that there are marked differences between the sexes in the predictive accuracy and this indicator is better in mandibles without molar tooth loss as mentioned by Loth & Henneberg. They also pointed out that the tooth loss is a potential source of error. On the other hand, in this study, there were no sexing errors when mandibles which ramus shape scores -1, 0 and +1 were extracted from the normative sample as the "sex indeterminate group". This study suggests that further assessment should be made only when both rami are available for observation, and the mandible shows either bilateral flexure (+2) or bilateral non-flexure (-2). The consideration of this new categorization can also be useful when using Loth & Henneberg's method. (c) 2004 Elsevier GrnbH. All rights reserved.