What is already known about this subject The index of body mass related to stature, (body mass index, BMI, kgm-2), is widely used as a proxy for percent body fat (%BF) in cross-sectional and longitudinal investigations. BMI does not distinguish between lean and fat mass and in children, the cross-sectional relationship between %BF and BMI changes with age and sex. What this study adds While BMI increases linearly with age from age 8 to 12 years in both boys and girls, %BF plateaus off between 10 and 12 years. Repeated measures in children show a systematic decrease in %BF for any given BMI from age 8 to 10 to 12 years. Because changes in BMI misrepresent changes in %BF, its use as a proxy of %BF should be avoided in longitudinal studies in this age group. Background Body mass index (BMI, kgm-2) is commonly used as an indicator of pediatric adiposity, but with its inability to distinguish changes in lean and fat mass, its use in longitudinal studies of children requires careful consideration. Objective To investigate the suitability of BMI as a surrogate of percent body fat (%BF) in pediatric longitudinal investigations. Methods In this longitudinal study, healthy Australian children (256 girls and 278 boys) were measured at ages 8.0 (standard deviation 0.3), 10.0 and 12.0 years for height, weight and percent body fat (%BF) by dual-energy X-ray absorptiometry. Results The patterns of change in the means of %BF and BMI were different (P < 0.001). While mean BMI increased linearly from 8 to 12 years of age, %BF did not change between 10 and 12 years. Relationships between %BF and BMI in boys and girls were curvilinear and varied with age (P < 0.001) and gender (P < 0.001); any given BMI corresponding with a lower %BF as a child became older. Conclusion Considering the divergence of temporal patterns of %BF and BMI between 10 and 12 years of age, employment of BMI as a proxy for %BF in absolute or age and sex standardized forms in pediatric longitudinal investigations is problematical.