Abstract: Objective To retrospectively analyze the prognostic significance of mean amplitude of glycemic excursion (MAGE) in patients with severe burn. Methods A study involving 76 patients with severe burn admitted into Department of Burns of the Third People's Hospital of Wuxi City from January 2011 to August 2014, conforming to the inclusion criteria was conducted. Data of their demography, injury, and treatment were collected. Blood glucose level of patients was measured during the first 72 h after admission, and MAGE was calculated. (1) The patients were divided into survival group (n=46) and death group (n=30) according to the outcome within post injury day (PID) 90. The MAGE level of patients was compared between two groups. (2) The patients were divided into 3 groups by tertiles of MAGE within 72 h after admission, with 26 cases in low tertile group (MAGE<3.3 mmol/L), 27 cases in middle tertile group (with MAGE from 3.3 to 5.5 mmol/L), 23 cases in high tertile group (MAGE>5.5 mmol/L). The surviving curve was drawn using the Kaplan-Meier method to compare survival rates and surviving time of patients among the 3 groups within PID 90. Data were processed with t test, one-way analysis of variance, LSD test, chi-square test, and Fisher's exact test. The surviving curve was analyzed by the Log-rank test. Correlation was analyzed between demography, acute physiology and chronic health evaluation Ⅱ score, injury, treatment, sepsis, length of ICU stay, MAGE and death of patients using the univariate and multivariate Cox regression analysis. Receiver operating characteristic (ROC) curve of MAGE was drawn to predict death for 76 patients. Results Within 72 h after admission, the MAGE of patients in death group was (6.0±1.4) mmol/L, which was significantly higher than that in survival group [(3.5±1.2) mmol/L, t=9.219, P=0.004]. The survival rates of patients in low tertile, middle tertile, and high tertile groups within PID 90 were respectively 88.5% (23/26), 59.3% (16/27), and 30.4% (7/23), with significant differences among 3 groups (χ²=18.073, P<0.001). The surviving time of patients in low tertile group [(83±21) d] was significantly longer than that of middle tertile group [(63±34) d, P<0.05]. The surviving time of patients was longer in low tertile and middle tertile groups than in high tertile group [(46±37) d, with P values below 0.05]. Total burn area, sepsis, blood purification, and MAGE were independent risk factors of death (with hazard ratio respectively 4.324, 1.591, 1.886, 2.047; 95% confidence interval respectively 2.978-6.119, 1.005-1.657, 1.614-2.408, 1.852-3.161; P<0.05 or P<0.01). The area under the ROC curve of MAGE for predicting death for 76 patients was 0.870 (with 95% confidence interval 0.786-0.966, P<0.001), and 4.7 mmol/L was chosen as the optimal threshold value, with sensitivity of 86.7% and specificity of 89.1%. Conclusions The increase of MAGE in patients with severe burn is significantly associated with poor prognosis; controlling the glucose level fluctuation guided by measuring MAGE may be helpful in reducing mortality of patients.