To determine the effect of glycemic control on the growth velocity and several metabolic parameters of children with insulin-dependent diabetes mellitus (IDDM), 79 patients with IDDM, 45 females and 34 males with a mean chronological age of 8.4 +/- 3.0 years were followed over a 5-year period starting at the onset of diabetes. Glycemic control was assessed by measuring total glycosylated hemoglobin; children were divided into better controlled, GHb < 9%, 30 children (Group A) and worse controlled, GHb > or = 9%, 49 patients (Group B). Growth velocity was significantly lower, in the five years of follow up, in the worse controlled patients when compared to the better controlled subjects (4.8 +/- 1.6 vs 6.7 +/- 2.2 cm/yr after the first year and 5.0 +/- 2.0 vs 6.5 +/- 1.8 cm/yr after the fifth year, in group B and group A, respectively). Higher cholesterol (185.3 +/- 33.7 vs 158.8 +/- 39.5 mg/dl) and triglyceride levels (85.9 +/- 43.5 vs 71.0 +/- 37.4 mg/dl) were apparent in the worse controlled patients, when compared to the better controlled children. Insulin dose was not significantly different in the two groups (0.76 +/- 0.3 vs 0.84 +/- 0.4 U/kg/day in the 1st year and 0.9 +/- 0.3 vs 0.92 +/- 0.4 U/kg/day in the 5th year, in group B and A respectively). Although both groups received the same initial and long term training by our pediatric diabetes team, more frequent blood glucose monitoring, better record keeping and rotation of injection sites and more clinic visits were clearly noted in the better controlled group. Ketoacidotic episodes were more common in the worse controlled patients, while better controlled children had a higher number of hypoglycemic episodes. In conclusion, we have found poor glycemic control, as reflected by higher glycosylated hemoglobin levels, to affect the growth velocity and several metabolic parameters of children with diabetes followed for a five-year period. Other factors besides insulin dose and initial and subsequent diabetic education seem to play a role in their glycemic control.