Body composition methods can be classified into direct, indirect and doubly indirect methods. In vivo direct methods use neutron activation analysis to get information on body composition. Indirect methods rely on rules and constants derived from direct methods. Most basic research, especially the development of rules and models has been done in Caucasian subjects in Europe or USA. The critical use of more advanced body composition methodologies in various ethnic groups has shown that assumptions may differ between ethnic groups, an example being the assumption of constant density of the fat free mass. Indirect or predictive methods rely on statistical relationships between body parameters and components of body composition. Subcutaneous fat patterning differs among ethnic groups, and this may have consequences for the validity of body fat predicted from skinfold thickness. Relative leg length and relative arm length also differ between ethnic groups. As a result the body mass index (weight/height squared, BMI), often used as surrogate for body fat percent, and formulas based on bioelectrical impedance measurement show different validity among ethnic groups. Less information is available about the validity of indicators for body fat distribution. There are indications that the relationship between the amount of visceral adipose tissue and waist circumference or waist-hip circumference ratio also differs among ethnic groups. Ethnic differences in body composition rules and constants are important and challenging to investigate, especially in relation to overweight and obesity.