Although continuous subcutaneous insulin infusion (CSII) has been used in pediatric practice for >20 yr, the technique is not widely used in many countries. The aim of this non-randomized population-based study was to evaluate CSII in routine pediatric care. In a 1-yr cross-sectional evaluation, 27/89 patients (30.3%, age 7-21 yr) used pumps (two during the night only), the others 4-6 injections/day. In patients with >2 yr of diabetes, pump users had higher HbA1c (8.9+/-1.0 vs. 8.2+/-1.6%, p=0.04), less insulin/24 h (0.9+/-0.1 vs. 1.0+/-0.2 U/kg, p=0.002), and longer diabetes duration (p=0.02). The higher HbA1c is explained by 67% of pump patients having high HbA1c (>8.5%) as the major indication for CSII. The overall incidence of severe hypoglycemia was 31.5/100 patient years, 40.3 for injection therapy, and 11.1 for pump therapy (p=not significant). The incidence of severe hypoglycemia with unconsciousness was 12.9/100 patient years and with seizures 9.7 for injection therapy, whereas no children on pumps experienced these complications during the cross-sectional study year. We had no admissions for ketoacidosis in either group during this year. The pump patients were followed for 5 yr after pump start. Two stopped using the pump after 2 and 3 yr. For the patients with high HbA1c as indication, mean HbA1c the year before pump was 9.5%. Mean HbA1c during the first year with pump was lowered to 8.9% (p=0.019), the second year 8.6% (p=0.017), the third year 8.6 (p=0.012), the fourth year 8.7 (p=0.062), and the fifth year 8.9% (p=0.28). We found six cases of ketoacidosis corresponding to 4.7/100 patient years. In conclusion, we found a long-term lowering of HbA1c after starting CSII in a pediatric population, decreased frequency of severe hypoglycemia, and a low risk of ketoacidosis.