To evaluate the efficacy of low-dose insulin therapy in cases of severe diabetic ketoacidosis (DKA), we examined admission clinical and biochemical parameters and responses to therapy in 48 diabetic patients who presented with DKA and were randomized to receive either high- or low-dose insulin. There were no differences in the initial clinical and biochemical parameters of the patients, regardless of assignment to low or high dose; however, a subgroup of 13 patients who were classified as severe DKA (based on their presentation in a comatose or stuporous state) had, as expected, more marked clinical and biochemical abnormalities than their alert cohorts. The responses to therapy (rate of glucose decrement and control of acidosis) were comparable in the high-dose and low-dose groups of comatose/stuporous patients and were not significantly different from the noncomatose cohorts. These data indicate that low-dose insulin therapy in severely ill comatose patients is as effective as high-dose.