Objective To evaluate the accuracy of simple anthropometric parameters in diagnosing obesity in children in Guangzhou. Methods A cross-sectional study, including 465 children aged 6-9 years, was carried out in Guangzhou. Their body height and weight, waist circumference (WC) and hip circumference were measured according to standard procedure. Body mass index (BMI), waist to hip ratio (WHR) and waist-to-height ratio (WHtR) were calculated. Body fat percentage (BF%)was determined by dual-energy X-ray absorptiometry. Multiple regression analysis was applied to evaluate the correlations between age-adjusted physical indicators and BF%, after the adjustment for age. Obesity was defined by BF%. Receiver operating characteristic (ROC) curve analyses were performed to assess the diagnostic accuracy of the indicators for childhood obesity. Area under-ROC curves (AUCs) were calculated and the best cut-off point that maximizing'sensitivity+specificity-1'was determined. Results BMI showed the strongest association with BF% through multiple regression analysis. For'per-standard deviation increase' of BMI, BF%increased by 5.3%(t=23.1, P<0.01) in boys and 4.6%(t=17.5, P<0.01) in girls, respectively. The ROC curve analysis indicated that BMI exhibited the largest AUC in both boys (AUC=0.908) and girls (AUC=0.895). The sensitivity was 80.8% in boys and 81.8% in girls, and the specificity was 88.2%in boys and 87.1%in girls. Both the AUCs for WHtR and WC were less than 0.8 in boys and girls. WHR had the smallest AUCs (<0.8) in both boys and girls. Conclusion BMI appeared to be a good predicator for BF%in children aged 6-9 years in Guangzhou.