The effect of low-dose intramuscular insulin therapy was compared with that of high-dose insulin therapy by intravenous and subcutaneous routes in 48 patients with diabetic ketoacidosis. A simplified protocol was devised to compare efficacy of the two methods of therapy in a randomized manner. Plasma glucose dropped to less than 250 mg/dl in the low-dose group in 6.7 +/- 0.8 h and in the high-dose group in 4.5 +/- 0.8 h (P = not significant). The amount of insulin necessary to lower plasma glucose to 250 mg/dl was 263 +/- 45 U in the high-dose group and 46 +/- 5 U in the low-dose group. Twenty five percent in the high-dose group and none in the low-dose group developed hypoglycemia. Other biochemical and clinical variables in the two groups were comparable. No treatment complications were noted in the low-dose group. Our studies suggest that low-dose intramuscular insulin therapy is simple and as effective as high-dose therapy in the treatment of diabetic ketoacidosis without the risk of hypoglycemia and with a diminished incidence of hypokalemia. Furthermore, the favorable response of these patients to low-dose insulin therapy suggests the absence of insulin resistance in diabetic ketoacidosis.