OBJECTIVE: To evaluate the reproducibility of the measurement of% body fat by bipedal biometrical impedance analysis (BIA) compared with anthropometric measurements of adiposity in children and the correlations between these methods in children and adults.METHODS: A cross-sectional study in a total of 1080 adults and children enrolled in 1999 in the Fleurbaix-Laventie Ville Sant?I (FLVS II) population-based study in northern France. The reproducibility of anthropometrical and BIA methods was determined by a nested analysis of variance of repeated measurements by 2 investigators and a bipedal BIA device (Tanita TBF 310) in 64 pupils of two 5th grade classes. The correlation of BIA and anthropometric adiposity measurements with the unknown relative fat mass or volume of the body estimated by a latent adiposity variable (LAV) was established by the triads' method in 1080 subjects of the FLVS II cohort.RESULTS: The reproducibility was similar for the sum of skinfolds, waist circumference and BIA% fat measurements (intraclass correlation coefficients: 0.979-0.992). Correlation coefficient between BIA body fat% and the LAV was higher than 0.86 in all sex and Tanner stage related groups, and similar in children and adults, except in pubertal boys (0.76).CONCLUSION: With a high level of reproducibility, foot-to-foot BIA analysis provides a valuable measurement of total% fat for epidemiologic studies in children. However further studies are needed before extrapolating these results to overweight children.