The prevalence of overweight and obese children has doubled, and the incidence of type 2 diabetes in children (0-19 y) has increased 4-fold during the past several decades. As a result we can anticipate an increased number of metabolic studies in children. There are few data on measures of glucose metabolism in normal children, and virtually none relating to their reproducibility. The aims of this study were 1) to provide new data on energy expenditure and glucose, lipid, and protein metabolism in nonobese, healthy children and adolescents; 2) to evaluate their reproducibility; and 3) on the basis of these data, to perform power calculations for metabolic studies. Eight nonobese subjects (8-16 y) were studied on two occasions, preceded by 7 d of a diet with identical energy content and macronutrient distribution. Gluconeogenesis, measured by deuterium oxide, accounted for 50% of glucose production. Insulin sensitivity, measured by the labeled minimal model, averaged 4.9 x 10(-4) mL(mU x min)(-1). Glucose appearance rate was significantly higher (p < 0.01) in the children than in the adolescents. Furthermore, we demonstrated that for energy intake and expenditure, plasma concentrations of glucose and C-peptide, and rates of appearance of glucose and leucine, a 10% difference can be detected in fewer than five subjects with a power of 80% and a type I error of 5%. Insulin concentration, gluconeogenesis, insulin secretory indices, insulin sensitivity, and glucose effectiveness were more variable, but with the above power a difference of 25% could be detected in 7-11 subjects using a paired study design.