This study evaluated the accuracy of surrogate indexes of insulin sensitivity (SI) in children. Surrogates (homeostasis model assessment index of insulin resistance, quick insulin sensitivity index, and 40/insulin ratio index) were cross-sectionally investigated in 66 obese and lean children (17 Tanner stage I, 19 Tanner stage II-III, and 30 Tanner stage IV-V) as indexes of insulin resistance in comparison with the minimal model. The pubertal decrease in SI was found with the minimal model (-47%; P = 0.01), but not with surrogates, which were not correlated to SI. Baseline insulin (Ib) did not mirror the decrease in SI, did not significantly change when plotted against pubertal stage or age, and was not correlated to SI. Ib and surrogates were positively correlated with the body mass index. The disposition index, which quantifies the feedback between SI and insulin release, was widely scattered and decreased during puberty (P = 0.05). The specificity and sensitivity of surrogates as predictors of insulin resistance were poor (e.g. 81.1% and 30.7%, respectively, for the homeostasis model assessment index of insulin resistance). Thus, during puberty, surrogates are not accurate predictors of insulin resistance. Because reference methods are rather expensive and invasive, additional studies of alternative techniques for evaluating SI are needed to allow accurate measurement of insulin resistance in children.