Teeth, essential components of stomatognatrhic system, constitute an excellent material for paleontological, archaeological, anthropological, genetic and forensic researches by virtue of their easy accessibility, examination, registration and almost indestructible structure. Information regarding their size and characteristics has been extremely useful in determining sex and in the process of human identification itself. This survey was an odontometric analysis of 481 upper dental stone casts (243 from males and 238 from females), of patients aged between 18 and 60 years, assisted in one orthodontic clinic of Montevideo, Uruguay, consisting of measurements of upper mesiodistal diameter of canines and intercanine distance.Three sex classification techniques were compared, the maxillary canine index and two supervised classification methods (classification tree and recursive partitioned tree), seeking to verify which method was the most suitable for the determination of sex, in a Uruguayan sample. The models showed a performance of 52 %, 77 % and 69 % respectively. It was concluded that the statistical models of supervised classification allowed a determination of the sex in a more reliable way than the discrimination method.