AIM:The objectives of this study are: (1) to study the relation between body mass index (BMI), percentage-weight-for-height (PWH) and percentage body fat (PBF) in Singaporean Chinese children; (2) to assess the applicability of an international definition of obesity (the International Obesity Task Force (IOTF) BMI) as a screening tool to detect childhood obesity, as compared with the current Singapore population-specific definition using PWH. METHODS: A total of 623 Chinese children aged 6-11 y (321 males, 302 females) were recruited from a school by proportionate (40%) stratified random sampling. BMI and PWH were calculated from weight and height, while PBF was derived using leg-to-leg bioelectrical impedance analysis. The strength of association among the three indices of obesity was assessed using Spearman's correlation coefficient. Obese children were defined as those above the 95th percentile of PBF in each agegender-specific group. Sensitivity and specificity of IOTF-BMI cutoff values and PWH cutoff values were compared by testing their ability to correctly identify obese children. RESULTS: All three indices correlated well with one another (BMI:PWH r ¼ 0.83, BMI:PBF r ¼ 0.87, PWH:PBF r ¼ 0.76). Prevalence of obesity was lower using IOTF-BMI cutoffs (6.9%) than using PWH cutoffs (16.4%). The sensitivity and specificity of IOTF-BMI cutoff values were 75.0 and 96.0%, respectively, with sensitivity differing between boys (83.3%) and girls (66.6%) (P ¼ 0.35). In comparison, PWH cutoff values had higher sensitivity (91.6%) but lower specificity (86.6%), with no significant difference between the genders. CONCLUSION: IOTF-recommended BMI cutoff values had low sensitivity and may underestimate the local prevalence of childhood obesity. For screening purposes, we recommend that population-specific measures rather than international cutoff values be used.