Introduction. Hyperglycemia during hospitalization is a risk factor that can be managed in order to reduce mortality. Inspite of hyperglycemia, glucose variability also brings negative outcome to cells. Studies about glucose variability effect to mortality had been studied using many variables of glucose variability. Methods. Retrospective cohort study is done to 280 critical ill patient in ICU and HCU in Cipto Mangunkusumo Hospital who admitted to critical care between January 2012-August 2013. MAG change and glucose standard deviation are divided into 4 quartiles. Relationship between MAG change and glucose standard deviation are analyzed using Chi Square test. To control the confounders (MSOFA score, Charlson comorbidities index, hypoglycemia, and hyperglycemia), logistic regression is done. Results. Median of MAG change is 3.3 mg/dL/hour and median of glucose standard deviation is 37.63 mg/dL. Mortality proportion is higher in upper quartile of MAG change and glucose standard deviation compared to lower quartile. OR of upper quartile MAG change to ICU mortality is OR 4.26 (95% CI 1.98-9.15) and OR of upper quartile glucose standard deviation to ICU mortality is OR 2.78 (95% CI 1.35-5.71). These results are adjusted to MSOFA score, hypoglycemia, and hyperglycemia. In logistic regression test, fully adjusted OR are 3.34 (95% CI 1.08-10.31) and 0.90 (95% CI 0.28-2.88) for MAG change and glucose standard deviation, respectively. Conclusions. Mortality proportion of upper quartile of MAG change (>8.1 mg/dL/hour) is higher than lower quartile (<1.3 mg/dL). Mortality proportion of upper quartile glucose standard deviation (>59 mg/dL) is higher than lower quartile(<22.7 mg/dL), but the difference is not statistically significant.