AIM: To identify the best anthropometric predictor of the metabolic syndrome in children. METHODS: Screening performance was evaluated in a clinical setting. The study included 140 children: 72 non-obese and 68 with non-syndromal obesity. Body mass index (BMI), waist circumference and triceps/subscapular skinfolds ratio were used as predictor variables, and systolic blood pressure, diastolic blood pressure, glucose, uric acid, fasting insulin, triglycerides and HDL-C as metabolic syndrome variables. RESULTS: The areas under the receiver operating characteristic (ROC) curves were 0.849 (95% CI: 0.780, 0.919) for BMI, 0.868 (95% CI: 0.801,0.934) for waist circumference and 0.834 (95% CI: 0.757,0.910) for the triceps/subscapular skinfolds ratio. No statistically significant differences were found for the three areas under the ROC curves. The point on the ROC curve closest to 1 corresponded to the 65th percentile for BMI, to the 70th percentile for waist, and to the 40th percentile for the triceps/subscapular skinfolds ratio. CONCLUSION: Waist circumference seems to be the best predictor of children with the metabolic syndrome in paediatric clinical settings.