Background: Our objectives were to determine whether persistent hyperglycemia when compared with normoglycemia was predictive of outcome in the later stages of hospitalization in critically injured trauma patients. Methods: A prospective study was conducted on 896 consecutive trauma patients admitted to the intensive care unit during a 2-year period. Patients were stratified by serum glucose level on day 1 to day 28 (low = 0-139 mg/dL, medium to high = 140-219 mg/dL, and high = >220 mg/dL), age, gender, race, insulin dependent diabetes, obesity, and Injury Severity Score (ISS). Patients were further stratified by pattern of glucose control (all low, all moderate, all high, improving, worsening, highly variable. Outcome was measured by ventilator days, infection, hospital and intensive care unit length of stay, and mortality. Multiple variable logistic and linear regression models were used to determine level of significance. Results: Eighty-three percent were victims of blunt trauma. The majority (74%) were male, with a mean ISS of 26 ± 12. Hyperglycemia (moderate, worsening, and highly variable) in the first week was associated with significantly greater hospital and intensive care unit length of stay, ventilator time, infection, and mortality when controlling for age, race, gender, ISS, mechanism of injury, obesity, and insulin dependent diabetes (p < 0.03). However, hyperglycemia in later weeks was not associated with infection and only weakly associated with mortality when analyzed by the same model. When controlling for glucose levels in subsequent weeks, patients who were normoglycemic in the first week had a lower infection rate and were less likely to die even when controlling for age, ISS, and obesity (p < 0.05). Conclusions: Early euglycemia is associated with improved outcome and appears to be protective regardless of glucose levels in subsequent weeks. Further studies are warranted to determine the etiology of this protective effect.