Objective: The incidence of neuropathic respiratory dysfunction in the general population ranges from 3% to 19% and may be high as 33% in patients with unconsciousness and neurologic impairment. Respiratory failure carries a 6-12-fold may increase mortality risk. The reasons advanced for increased mortality include old age, the presence of persistent hyperglycemia, and the act of reintubation itself resulting in serious complications. Methods: DM patients with respiratory failure were divided into two groups: those given insulin therapy with control blood sugar around 110 ~ 120 mg/dL and blood sugar over 180 to 200 mg/dL. Then, descriptive and inferential statistics were analyzed and presented as means ± standard deviation for continuous variables and frequencies for categorical variables based on the two groups. The differences in demographic, clinical, and laboratory characteristics between the groups were examined using independent samples t tests. The associations between the variables and patient characteristics were analyzed using χ2 or Fisher exact tests. Results: Outcome analyses showed that the well blood sugar control group had a higher survival rate, but not associated of successful extubation, a shorter MV duration, shorter ICU and hospital stays, and lower medical costs. Moreover, a hierarchical regression model controlled for age; sex; APACHE II, TISS, and GCS scores; hemoglobin and albumin levels; comorbidities, showed that rehabilitation therapy was significantly negatively associated with MV duration, but not significantly associated with lengths of ICU or with medical expense. Conclusion: Moreover, the systemic inflammatory state can contribute to muscle dysfunction. There are several plausible mechanisms supporting the positive impact of adequate blood sugar control with acute respiratory failure in ICU. Severe complication of diabetes in ICU may cause severe physical inactivity can result in reduction in muscle protein synthesis, lean mass and leg muscle strength.