We evaluated the outcome of pregnancies followed between 1990 and 2000 in 93 women with type 1 diabetes, treated with conventional intensive insulin therapy (n=68) or continuous subcutaneous insulin infusion (n=25). We evaluated metabolic control (fasting and 1-hour post-prandial plasma glucose and HbA1c levels), spontaneous or induced abortions, time and mode of delivery, maternal outcome (pregnancy-induced hypertension, preeclampsia, placental insufficiency, hydramnios, hypoglycemic coma, ketoacidosis) and fetal outcome (weight, hypoglycemia, hypocalcemia, hyperbilirubinemia, fetal distress, asphyxia, hyaline membrane disease, polycythemia, shoulder dystocia, malformations). Patients treated with insulin pump more frequently had background retinopathy and clinical neuropathy. No significant differences were observed between the two groups in metabolic control and maternal outcome. Glycemic control, non-optimal in the prepregnancy state, improved significantly during pregnancy, as shown by the progressive reduction in HbA1c levels. As regards fetal outcome, no differences were observed between the two groups in morbidity and especially in malformation rate. Patients with malformed babies did not have optimal metabolic control at conception. Thus, maternal and perinatal outcomes were comparable in patients treated with insulin pump and continuous subcutaneous insulin therapy, and depended on metabolic control. In patients in higher White's class and with more unstable glycemia, we achieved metabolic control and outcomes comparable with those of women of lower White's class and more stable glycemic values using the insulin pump. Our data suggest that insulin pump therapy is useful in problematic, complicated cases of women who want a baby.