OBJECTIVES: The effect of diabetes and diabetic medications on the morbidity and mortality in hospitalized cancer patients is a serious concern. We examined the association of diabetic medications with the transfer to ICU and inpatient mortality. METHODS: Electronic patient records were prospectively collected in 5489 hospitalizations to UT MD Anderson Cancer Center from May 1st, to July 31st, 2006. For each patient demographic, laboratory and pharmacy data were obtained.  To observe the effect of diabetic medications, we selected only diabetic patients, which were 36.3% (1991) of the total hospitalization. Descriptive and logistic regression analyses were performed. RESULTS: Out of 1991 hospitalizations for diabetic cancer patients, fifty seven percent were male, sixty nine percent were White  with median age of sixty one. These patients were on different diabetic medications and  thirty one percent were only on sliding scale insulin (SSI).After controlling for socio-demographic variables and other medications like chemotherapy, we found cancer patients who were only on SSI were more likely to be transferred to intensive care unit (ICU), (OR 2.31, CI: 1.7-3.2; P<0.001) and die during hospitalization (OR 1.88, CI: 1.3-2.7; P<0.001). Glucocorticoids administration was also significantly associated with inpatients mortality (OR 5.91, CI: 3.5-9.8; (p<0.00)). CONCLUSIONS: We found strong associations between the method of administration of insulin (SSI) and transfer to ICU and inpatient mortality in cancer patients with diabetes. Drug dosage and administration in inpatient settings should be tailored to the need of the patients to optimize the medication effect and minimize the side effects.