The modified minimal model (MMM), a recently introduced method that assesses insulin sensitivity (SI) by a computed mathematical analysis of the relation between the change in insulin and glucose clearance after a bolus of iv glucose, followed 20 min later by a bolus of tolbutamide, has been standardized in adults, but this method has not been validated in children. We performed an abbreviated 90-min MMM test in 50 children who were siblings of patients with insulin-dependent diabetes mellitus and 7 healthy adult volunteers and compared the results to the standard 180-min MMM test in 11 of these subjects. The cohort consisted of 29 prepubertal children [16 males and 13 females; 8.7 +/- 2.0 (mean +/- SEM) yr old]; 16 pubertal children defined as less than 17 yr of age and Tanner stage 2-5 (8 males and 8 females; 13.4 +/- 1.8 yr old), and 12 postpubertal subjects (7 males and 5 females; 18.2 +/- 0.9 yr old), with no significant difference in the weight for length index (WLI) among the 3 groups and with sera of all subjects negative for islet cell antibodies and insulin autoantibodies. The test procedure consisted of 3 baseline blood samples over 30 min, followed at zero time by 0.3 g/kg 25% dextrose infused iv over 1 min and an iv injection of tolbutamide (5 mg/kg) 20 min later; sequential blood samples for glucose and insulin measurements were withdrawn from zero time until completion 90 or 180 min later. In the 11 subjects who underwent both the standard and the abbreviated tests, there was no significant difference between the SI estimated by the 2 methods provided that glucose and insulin values were interpolated at 180 min during the computer calculations of the abbreviated test. Using the 90-min abbreviated test, the SI of the pubertal subjects (2.92 +/- 0.45) was markedly less than that of the prepubertal subjects (6.57 +/- 0.45; P = 0.0001). While the postpubertal group value of 4.63 +/- 0.86 was significantly higher than that of the pubertal group (P = 0.0001), the pre- and postpubertal groups remained significantly different (P = 0.0001). The 10 obese subjects with WLI greater than 120% had a lower SI (3.5 +/- 0.53) than the 47 nonobese subjects with WLI less than 120% (SI = 5.48 +/- 0.42; P less than 0.04), and there was a negative correlation between SI and WLI. None of the study subjects experienced symptomatic hypoglycemia during the test.(ABSTRACT TRUNCATED AT 400 WORDS)