Objective: Dysglycemia is associated with poor outcomes; the actual status of dysglycemia of critically ill patients with diabetes should refer to background glycemia. We investigated the effect of difference between mean blood glucose and basic blood glucose upon outcomes.Methods: Glycated hemoglobin A1c (HbA1c) was detected within the first 24 h and converted to A1C-derived average glucose (ADAG) by the equation ADAG = [(HbA1c & lowast;28.7) - 46.7]& lowast;18-1; blood glucose measurements were fourth per day during the first 7 days after admission; the mean blood glucose level (Mean), standard deviation (SD), and coefficient of variation (CV) were calculated. GAP were calculated as admission blood glucose and Mean minus ADAG, respectively.Results: Six hundred forty-nine patients were recruited and 428 survived at 28 days; 302 patients with diabetes had greater ADAG, blood glucose at admission (BGadm), Mean, SD, CV, GAP, and hypoglycemia incidences. The GAP between Mean and ADAG had superior predictive power, which was decreased in patients with diabetes and increased in patients without diabetes. GAP7 was related to 28-day mortality; the death risk was decreased in patients with diabetes. Patients with lower GAP tended to survive. Nonsurvivors with diabetes suffered higher rate of hypoglycemia than survivors which was the opposite in patients without diabetes.Conclusion: The glycemic GAP between the mean level of blood glucose within the first 7 days in ICU and ADAG was independently associated with 28-day mortality of critically ill patients, which was different between patients with and without diabetes. Hypoglycemia in patients with diabetes should be a concern.