There are limited data on which to base insulin dosing schemes for diabetic ketoacidosis (DKA). The goal of therapy is to avoid excessive decreases in serum glucose (greater than 100 mg/dL/h) because of the risks of rapid changes in serum osmolarity and the potential risk of cerebral edema. We retrospectively reviewed the therapy of DKA in pediatric patients admitted to our Pediatric Intensive Care Unit over the past 10 years. There were 35 patients who received IV bolus insulin therapy (0.08 to 1.6 units/kg, 0.24 +/- 0.27 units/kg). The serum-glucose decrease was less than or equal to 100 mg/dL in 10 patients, 101 to 200 mg/dL in 13 patients, 201 to 300 mg/dL in 8 patients, 301 to 400 mg/dL in 2 patients, and more than 500 mg/dL in 2 patients. In patients who received 0.05 to 0.1 units/kg of insulin as a bolus dose, the decrease in serum glucose was greater than 100 mg/dL in 5 of 11 patients. An insulin infusion was administered to 91 patients. During the 243 hours of insulin infusion therapy, the decline in serum glucose was 0 to 100 mg/dL during 162 hours, 101 to 200 mg/dL during 49 hours, 201 to 300 mg/dL during 8 hours, and more than 300 mg/dL during 3 hours. Of the 193 hours of 0.05 to 0.1 units/kg/h insulin administration, there were 47 hours (24%) during which the serum-glucose decrease was greater than 100 mg/dL. Of the 21 hours of insulin administration at less than 0.05 units/kg/h, there was 1 hour (4.8%) where the serum-glucose decrease was greater than 100 mg/dL (P = 0.05 vs. insulin infusion at 0.05 to 0.1 units/kg/h). Commonly used insulin dosing regimens of a bolus of 0.1 units/kg followed by an infusion of 0.05 to 0.1 units/kg/h frequently resulted in a decrease in serum glucose of greater than 100 mg/dL/h. Prospective trials are needed to more accurately define appropriate insulin dosing regimens for pediatric patients with DKA.