Correct race and sex determination of unknown skeletal material is an important aspect of forensic anthropology. Numerous studies have focused on the differences, both osteometric and morphological, between the sexes of a particular racial phenotype, between race groups, and populations. From previous work by a variety of researchers, the necessity of population specific standards for identification has been demonstrated. The purpose of this research was to examine the metric and morphological differences in the pelvis between the sexes and races of South African whites and blacks. Results will be used in developing standards of identification tailored to this population. A sample of 400 known sex/race os coxae were examined. Skeletal material was obtained from the Pretoria collection housed at the University of Pretoria, Department of Anatomy and the Dart collection located at the University of Witwatersrand, Department of Anatomical Sciences. A series of thirteen measurements and five morphological characteristics were examined. Indices were calculated from data obtained from the metric analysis. Left and right sides were examined and those bones visibly pathologically deformed were excluded from the study. Data were subjected to SPSS stepwise and direct discriminant analysis. Results showed ischial length as the most sexually dimorphic characteristic in whites, while acetabulum diameter was best in blacks. Four functions (using pelvic dimensions) were developed for determining sex. Highest accuracies were achieved from function 1 (including all dimensions) which correctly classified 92-96% of individuals. Race differences were also investigated. Pubic length was chosen as best for discriminating between races for males and iliac breadth as best in females. Accuracies were 86-89% for males and 82-88% for females. Accuracies for sex discrimination were consistent with earlier studies. Morphological results yielded >80% accuracy for all traits in white males except greater sciatic notch shape where only 33% were correctly classified. A population specific variation in sciatic notch shape was observed where >50% of the white males had a wide sciatic notch previously thought to be a female expression. Black males recorded 81 % correct classification for pubic shape and >90% for the remaining characteristics. Greater sciatic notch and pubic bone shape achieved highest accuracies with 96% for both traits in white females, and 84% and 88% in black females respectively. In conclusion, this study conclusively demonstrates that race and population differences affect the expression of sexual dimorphism and must be accounted for to develop the most effective methods of analysis.