Abstract Sex determination using mandible parameters is population dependent. In order to assess which measurements better characterize sex in prehispanic individuals from the Canary Islands, we blindly contrasted the results obtained by visual inspection and osteometric measurements with those obtained by molecular sexing using amelogenin ancient DNA analysis on teeth from the same material. Unambiguous sex classification was achieved by amplification of sex specific amelogenin alleles in 56 out of 76 mandibles (73.78% of the cases). Visual inspection led to a correct diagnosis in 66.04% of cases, with a greater proportion of errors for female (54.17%) than male (17.24%) mandibles. Osteometric measurements were able to assign sex correctly in 72.2% in the best of cases (mandibular height), a proportion similar to that obtained using a discriminant function (71.2%). By logistic regression analysis, ramus breadth, index ramus breadth/ramus height and mandibular length were the parameters independently related with a mistaken diagnosis of female sex, whereas bigonial width, ramus height and mandibular length were the parameters more closely and independently related to a mistaken diagnosis of male sex. In conclusion, diagnosis based on visual examination of the mandible or on its metric measurement only serves to roughly estimate sex with an accuracy of around 70% or less, at least among the prehispanic population from Gran Canaria. Amplification of amelogenin alleles leads to unambiguous identification of male and female alleles in 73.68% of cases, at least among the prehispanic population from Gran Canaria.