This study defines determinants of cost of treatment of diabetic ketoacidosis. A chart review for 92 cases of ketoacidosis from 1988 to 1992 in 40 females and 25 males (age range 18 to 81) showed a length of stay of 5.7 +/- 4.9 d. Length of stay did not correlate with the level of nursing care or any other component of the initial order set. Length of stay was shorter in cases managed by diabetologists. Length of stay was inversely proportional to the interval from arrival to the first shot of intermediate or long-acting insulin. Length of stay was longer in subjects with a positive bacterial culture (mean, 9.1 d) and was longer in subjects who arrived in the evening. There was a female predominance in total and recurrent cases of ketoacidosis. Female patients received fewer educational sessions than males. The grade of acidosis affected the duration of acidemia, but the grade of acidosis, APACHE scores, and admission lab values did not correlate with length of stay. The use of an intensive care unit (ICU) included more testing and expense without uniform clear benefit. Optimal transition from intensive to routine management includes resumption of long-acting insulins as soon as possible. Optimization of hospital care and reduction of incidence of ketoacidosis in females would have a marked effect on health care costs.