We studied 58 children with diabetic ketoacidosis using a random, prospec tive protocol, with insulin administered either as a low-dose continuous in fusion or as high-dose intermittent subcutaneous injections. There were no statistically significant differences between admission pH and glucose deter minations or the time to metabolic correction. The incidence of hypoglycemia and hypokalemia was higher in patients receiving subcutaneous insulin. In sulin levels in the low-dose patients were 85-160 μU/ml. The insulin required to achieve metabolic recovery was 1.6 U/kg in the low-dose group and 4.5 U/kg in the high-dose group (p < 0.01). Glucose administered at a rate of 3 to 4 g per unit of insulin infused in the low-dose group maintained a serum glucose of 150 to 250 mg/dl. Our studies suggest that low-dose intravenous insulin therapy is safe, as effective as high-dose intermittent subcutaneous injections and avoids the risks of hypoglycemia and hypokalemia. Meticulous attention to individual patient care, however, must remain the most important single variable.