Sexual differences in the human skeleton have been well studied in many populations. Odontometric analysis of the human sexual variation has been less investigated and mostly derived from the dentition of extinct populations. Turkey is situated in a unique location where populations from different regions mixed with each other and created a rich gene pool. One might anticipate that modem Turkish population is composed of genes from the Balkans, Caucasus, Middle East, Iran and further as well as from ancient Romans, Byzantines, Arabs and Asiatic Turks. It is clear that contemporary Turks are a mixture of these extant and extinct people and ideal to consider it a representative study population. The purpose of this study is to analyze dental dimensions and sexual variation in living Turks and develop forensic techniques to identify human remains from the teeth when any other technique is not available or not reliable. The study is composed of Ankara University dental students (50 male and 50 female casts, average age of 21 years). Bucco-lingual breadths from 14 teeth (11 through M2 of the maxilla and mandible) are taken from the left side and analyzed using the discriminant function statistics. An intraobserver error test did not indicate any statistically significant difference between any two measurements. Results of the study revealed that males exceeded females significantly (P < 0.001) in dimensions. Coefficient of variation was most obvious in I1s and I2s of both jaws in both sexes. Stepwise discriminant function statistics suggested that upper C, and lower C and M2 are the most contributory teeth to the function. Additional formulae were calculated for situation in which only one or a fragmented jaw is available for identification. Overall accuracy of sex diagnosis ranged from 73 to 77%. In conclusion this research supports earlier studies that sexual dimorphism is population specific. While dental difference between the sexes in several human populations has been found highly dimorphic, it was not found so in Turks and accuracy of classification remained low at about 77%. The difficulty or the lack of dimorphism comes from male subjects. (C) 2003 Elsevier Ireland Ltd. All rights reserved.