BACKGROUND: Hyperglycemia is associated with complications in the surgical intensive care unit. The purpose of this study was to determine the efficacy and safety of nurse-driven insulin infusion protocols in lowering blood glucose (BG) in critical illness. STUDY DESIGN: All patients in a 24-bed surgical intensive care unit who required IV insulin infusions during 3 noncontiguous 6-month periods from 2002 to 2004 were evaluated. In the preintervention phase, 71 patients received a physician-initiated insulin infusion without a developed protocol. They were compared with 95 patients who received a nurse-driven insulin infusion protocol with a target BG of 120 to 150 mg/dL and to 119 patients who received a more stringent protocol with a target BG of 80 to 110 mg/dL. RESULTS: There was a stepwise decrease in average daily BG levels, from 190 to 163 to 132 mg/dL (p < 0.001). The less stringent protocol decreased the time to achieve a BG level < 150 mg/dL from 14.1 to 7.4 hours compared with physician-driven management (p < 0.05) resulting in similar time on an insulin infusion (53 versus 48 hours). The more intensive protocol brought BG levels < 150 mg/dL in 7.2 hours and < 111 mg/dL in 13.6 hours, but increased the length of time a patient was on an insulin infusion to 77 hours. The incidence of severe hypoglycemia (BG < 40 mg/dL) was statistically similar between the groups, ranging between 1.1% and 3.4%. CONCLUSIONS: Implementation of a nurse-driven protocol led to more rapid and more effective BG control in critically ill surgical patients compared with physician management. Tighter BG control can be obtained without a significant increase in hypoglycemia, although this is associated with increased time on an insulin infusion.