Objectives: Despite evidence suggesting caution, employment of body mass index (BMI, kg m−2) as a proxy for percentage of body fat (PFat) in longitudinal studies of children and adolescents remains commonplace. Our objective was to test the validity of change in BMI as a proxy for change in PFat measured by dual-energy X-ray absorptiometry (DXA) during adolescence. Design: Longitudinal study. Methods: Healthy, predominantly Australian youth of mainly Caucasian background (131 females and 115 males) underwent repeated measures at 12.0 (SD 0.3) and 16.0 (SD 0.3) years for height, weight and PFat (DXA). Results: There was no significant difference in the percentage changes in BMI and PFat for the females (β = 2.45, standard error (SE) = 1.39, 95% confidence interval (CI) = [−0.27; 5.17]) with their mean BMI increasing 15% as their mean PFat increased 18%. However, for the males, while their mean BMI also increased 15%, their mean PFat was reduced 25%; this change being highly significant (β = −42.25, SE = 2.23, 95% CI = [−46.22, −38.27]). Conclusions: While change in BMI is likely to be a rough proxy for change in PFat measured by DXA in longitudinal studies of adolescent females, this is not the case for adolescent males, where increased BMI is likely to correspond with decreased PFat. Consequently, inferences from longitudinal studies of adolescents which have assumed that an increase in BMI (or BMI Z-scores or percentiles) represents an increase in adiposity require reconsideration.