Patients with acute stress from stroke or myocardial infarction may develop hyperglycemia, even in the absence of diabetes mellitus. Stress hyperglycemia increases morbidity and mortality in these patients. There has been no study to evaluate stress hyperglycemia in patients with sepsis.A prospective cohort study in patients with sepsis admitted in the Department of Medicine of Siriraj Hospital during 2006-2007 was done. Data were collected on admission blood glucose, HbA1c and other factors which may predict the outcomes of sepsis.Data were collected from 70 patients with sepsis. The prevalence of stress hyperglycemia was 42.3% in this study. We found no differences in clinical findings, laboratories, interventions and outcomes between groups of stress and non-stress hyperglycemia. Multivariate analysis showed that only APACHE II score and use of a mechanical ventilator were associated with mortality.The prevalence of stress hyperglycemia in patients with sepsis was high. We cannot conclude that stress hyperglycemia did not affect the mortality and morbidity outcome mainly because of the small number of subjects which may be not enough to detect statistical significance.