Identification of sex from the skeleton is an important demographic assessment in medicolegal investigations. Studies have demonstrated that populations differ from each other in size and proportions and that these differences can affect the metric assessment of sex. It is therefore vital to determine if population differences are great enough to necessitate group-specific standards. To date, there have been no attempts to create standards of assessment for modern Thais. Therefore the purpose of this research is to establish standards from which to determine sex from the femur using a new skeletal collection housed at the Chiang Mai University Department of Anatomy. The sample is composed of 104 individuals (70 males, 34 females). Six standard osteometric dimensions including maximum length, maximum head diameter, midshaft circumference, midshaft anterior-posterior and transverse diameters, and bicondylar breadth were measured and analyzed by stepwise discriminant function statistics. To understand population differences, formulas derived from Chinese, South African whites and American whites and blacks using the same method and variables were tested on the Thai sample. Results indicated that maximum head diameter and bicondylar breadth are the optimal combination for sex diagnosis and yielded 94.2% accuracy. Direct analysis using predetermined single or multiple variables also revealed bicondylar breadth as the best single dimension (93.3%). In cross-tests on the Thais, the Chinese formula gave the most favorable outcome with unsatisfactory results for all other groups. The present research confirms that sexual dimorphism is better reflected in breadth dimensions than in bone length. Comparisons showed that Thais are very different metrically from whites and blacks, and although they most resemble the Chinese, these two groups are not identical. These findings underscore the need for population-specific formulas for identification of sex from the skeleton.