The aims of this study were to quantify the effect of obesity definition on estimates of prevalence, and to determine the sensitivity and specificity of three commonly used definitions of overweight/obesity in children: body mass index (BMI) s.d. score > 2.00; BMI s.d. score > 1.04; weight > 120% ideal. A representative community sample of children in Edinburgh, Scotland (n=240, 124 boys and 116 girls; mean age 8.5 s.d. 0.4y) was recruited. Obesity was defined by a criterion method based on % body fat: > 25% fat in boys; > 32% fat in girls. Sensitivity of BMI s.d. score > 2.00 was relatively poor in both sexes (60% in girls; 36% in boys) but had high specificity (98%). Sensitivity of the other two clinical definitions was higher, and was better in girls than boys, but with lower specificity. Choice of definition had a profound effect on prevalence estimates. In conclusion, sensitivity of the definitions of obesity currently recommended for children, when tested in this sample, was heavily dependent on the definition used and differed between boys and girls. This should be considered when choosing a definition of obesity in clinical practice and epidemiology.