OBJECTIVE Insulin resistance increases during adolescence, and is exaggerated in patients with insulin dependent diabetes mellitus (IDDM). A relative deficiency of insulin-like growth factor-I (IGF-I) may contribute to this increased insulin requirement. Two mechanisms have been proposed: (a) increased GH secretion, caused by failure of IGF feedback control, leading to increased insulin resistance and (b) lack of insulin-like action of the IGFs which is reinforced by high plasma levels of IGFBP-1, an inhibitor of IGF action. The contribution of these two mechanisms to the 'dawn phenomenon' is assessed. METHODS The two possible mechanisms were studied during the dawn rise of glucose in pubertal adolescent patients with IDDM. Two overnight studies were performed in each subject. Patients remained on the same insulin regimen throughout. METHODS Twenty-two diabetic adolescent subjects, aged (mean +/- SEM) 14.0 +/- 0.4 years, duration of IDDM 7.9 +/- 0.8 years, were recruited. Pubertal status was: group 1 (breast stage 1-2; testicular volume < 4-8 ml) 3 male and 4 female, group 2 (breast stage 3; testicular volume 10-12 ml) 0 male 4 female, group 3 (breast stage 4-5; testicular volume 15-25 ml) 4 male and 7 female. Height standard deviation score (mean +/- SD) (-0.02 +/- 0.99) and daily insulin dose (50.4 +/- 3.1 U/day) did not change between studies. There were no differences in HbA1 (study A 11.26 +/- 0.45%, study B 11.09 +/- 0.42%). METHODS The subjects were admitted for the two studies 0.3 +/- 0.03 years apart. Blood samples were taken via an indwelling cannula every 20 minutes between 1900 and 0700 h. METHODS GH was assayed every 20 minutes, IGFBP-1, glucose and free insulin every hour and IGF-I at 0700 h. GH, IGFBP-1, IGF-I and free insulin were measured by radioimmunoassay. IGFBPs were also analysed by Western ligand blotting techniques. GH profiles were analysed by Pulsar and results compared by paired Student's t-test. The relations between the dawn rise in glucose and the changes in IGFBP-1, GH and free insulin were examined by multiple linear regression analysis. RESULTS Serum IGFBP-1 levels rose overnight in the two studies (study A, from 9 +/- 1 at 2200 to 59 +/- 9 micrograms/l at 0700 h; study B, from 10 +/- 1 at 2100 to 64 +/- 14 micrograms/l at 0700 h) whilst insulin levels fell from 47 +/- 5 at 2200 to 16 +/- 2 mU/l at 0700 h (study A) and from 45 +/- 5 at 2000 to 14 +/- 2 mU/l at 0700 h (study B). Glucose levels fell from 16.0 +/- 1.0 to 9.3 +/- 0.9 mmol/l at 0400 h, and then rose to 11.9 +/- 1.1 mmol/l at 0700 h during study A, and from 13.4 +/- 1.3 to 10.1 +/- 1.1 mmol/l at 0400 h and then rose to 13.5 +/- 1.0 mmol/l at 0700 h during study B. There were no differences in GH secretion between studies (mean GH levels (mean +/- SD) (study A, 15.7 +/- 6.6 mU/l; study B, 16.2 +/- 7.1 mU/l; correlation within subjects between studies r = 0.77, P < 0.001), sum of GH peaks (study A, 189.9 +/- 90.3 mU/l; study B, 185.8 +/- 100.2 mU/l; r = 0.57, P = 0.006)). Mean GH levels varied with pubertal stage (group 1, 12.1 +/- 1.5 mU/l; group 2, 23.3 +/- 2.1 mU/l; group 3, 15.3 +/- 1.2 mU/l). Serum IGF-I levels were not different (study A, 203 +/- 12 micrograms/l; study B, 218 +/- 13 micrograms/l). REGRESSION ANALYSIS: The change in plasma glucose between 0200 and 0700 h in both studies related to free insulin, IGFBP-1 and the sum of the GH levels over the preceding hour (log glucose = 7.87 + 5.32 log IGFBP-1 (P = 0.0001) - 5.05 log free insulin (P = 0.0001) - 1.44 log GH (P = 0.004); R2 = 72%). Mean overnight GH levels did not predict the morning rise in plasma glucose. CONCLUSIONS The morning rise of IGFBP-1 and plasma glucose appear to be related in this group of subjects with IDDM and this was a consistent finding in the two studies. This relation was additive to the effect of insulin deficiency.