Mesiodistal and buccolingual measurements, commonly used as a means of sex determination from teeth, can sometimes cause difficulties. The aim of this study is to test whether diagonal measurements can make it possible to take more accurate measurements. The results of diagonal measurements of dental casts taken from 30 males and 30 females have been evaluated by discriminant function statistics. Intra- and interobserver error tests did not indicate any statistically significant differences between the findings of two observers. Seven of the 14 measurements on the maxilla and 10 of the 14 measurements on the mandible have been found to be significantly greater in males. According to the results of the stepwise discriminant function statistics, the most contributory measurements to the function were upper first incisor mesiobuccal–distolingual (MBDL) and distobuccal–mesiolingual, lower second incisor MBDL, and lower canine MBDL. The highest reliability was obtained in MBDL measurements. It was realized that diagonal measurements of teeth, especially of canines, revealed clear dimorphic differences. Classification accuracy was found to be 83.3% for total sample, 78.3 for upper jaw, and 85.0% for the lower jaw. Accuracy rate was higher in the lower teeth. Commonly seen orthodontic anomalies, such as tooth rotations, crowding, attritions, deep dentin–enamel junction defects, and certain types of fillings, could make it difficult to correctly take width measurements or could cause other mistakes to occur. This explains the reason why this research has been considered to be of some use in diagonal measurements, which is an accurate method, particularly when employed for the front teeth.