Résumé : Objectives. To investigate cortisol levels in brain dysfunction in patients with severe sepsis and septic shock. Methods. In 128 septic and sedated patients, we studied brain dysfunction including delirium and coma by the evaluation of Richmond Agitation Sedation Scale (RASS), the Confusion Method Assessment in the ICU (CAM-ICU) after sedation withdrawal and the measurement of serum S100B biomarker of brain injury. Serum cortisol and S100B were measured within 12 hours after ICU admission and daily over the next four days. Results. Brain dysfunction was observed in 50% (64/128) before but in 84% (107/128) of patients after sedation withdrawal, and was more common in the patients older than 57 years (P = 0.009). Both cortisol (P = 0.007) and S100B levels (P = 0.028) were higher in patients with than patients without brain dysfunction. Cortisol levels were associated with ICU mortality (hazard ratio = 1.17, P = 0.024). Multivariate logistic regression showed that cortisol (odds ratio (OR): 2.34, 95% CI (2.01, 3.22), P = 0.02) and the combination effect of cortisol with age (OR: 1.004, 95% CI (1.002, 1.93), P = 0.038) but not S100B were associated with brain dysfunction. Conclusions. Cortisol was an associated-risk factor of brain dysfunction in patients with severe sepsis and septic shock. © 2014 Duc Nam Nguyen et al.