The efficacy and feasibility of the insulin infusion pump in pregnancy were examined in seven class D to FR diabetics, maintained on the pump at home from 10 to 29 wk gestation until delivery. An improvement in glucose control was achieved within the first month and sustained to term. Home monitoring demonstrated a fall in mean blood glucose levels from 135 mg/dl range 98-175, prepump) to 104 mg/dl (range 84-120) and a 25-30% reduction in standard deviations during pump treatment. Furthermore, glycosylated hemoglobin levels were normalized in each patient. During periodic inpatient evaluations, mean 24-h plasma glucose levels were slightly, but not significantly, lower after pump treatment (97 vs 86 mg/dl). However, glycemic excursions were strikingly diminished; MAGE values and standard deviations fell by 45% and 34%, respectively. The total daily insulin dose required at the start of pump treatment was 31% less than the conventional dose used before the pump (P less than 0.002). Thereafter the insulin dose increased by approximately 2.5 U/wk, with the basal infusion remaining at 40% of the total dose throughout the pregnancy. All infants were born at term, had no macrosomia or neonatal problems, and had normal intravenous glucose tolerance tests at 2 h of age. We conclude that the insulin infusion pump, managed at home, is a highly efficient way to achieve normal or near-normal glucose levels in the pregnant diabetic.