One hundred and forty episodes in 112 patients (58 boys) with diabetic ketoacidosis (96 episodes) and diabetic ketosis (44 episodes) were studied to elucidate the clinical and laboratory risk factors for altered level of consciousness at presentation and to analyze the outcome of a distinct protocol in the treatment of diabetic ketoacidosis. The patients were analyzed according to demographic data and clinical and laboratory findings at admission. The treatment protocol involved use of 0.45% sodium chloride (NaCl) in 2.5% dextrose as the initial fluid therapy following volume expansion. Dextrose content of the fluid was doubled once the serum glucose level fell below 250 mg/dl. The mean ages at presentation with diabetic ketoacidosis and ketosis were 10.3 +/- 4.4 and 10.2 +/- 4.0 years, respectively. Thirty-one percent of patients had altered consciousness at presentation. The level of consciousness correlated negatively with serum bicarbonate level (r=-0.485; p<0.001). A serum bicarbonate level below 15 mmol/L was a risk factor for altered consciousness. There was no correlation between effective osmolality and the level of consciousness. Serum effective osmolality above 320 mOsm/kg H2O did not appear to be a risk factor for altered consciousness. No mortality or any signs of clinical brain edema were observed in patients treated with the distinct treatment protocol. In conclusion, acidosis appears to be the major factor in the pathogenesis of altered consciousness at presentation. Serum effective osmolality does not seem to be a risk factor as suggested previously. Dextrose added to the infusion fluid early in treatment seems to prevent the development of brain edema, and this may be due to a protective effect of higher osmolality in the resultant solution.