Moderate hypothermia may reduce subsequent neuronal damage after traumatic brain injury. Interleukin (IL)-6 may have a role in the pathogenesis of traumatic neuronal damage or repair. Using the enzyme-linked immunological sorbent assay (ELISA), we serially measured IL-6 levels in plasma obtained from the radial artery (systemic) and internal jugular vein (regional) in 13 cerebral trauma patients who underwent hypothermia of 32-33 degrees C ranged from 4-9 days postinjury and 10 head-injured patients who were maintained at normothermic levels (36-37 degrees C). In both patient populations, surface cooling was used since even in the normothermic group, cooling was needed to maintain patient temperature in the normothermic range. All patients were mechanically ventilated after injection of midazolam and vecuronium. The administration of these agents were continued until the end of the study. Hypothermia was typically maintained for four days, however, in some cases based upon CT findings and/or intra-cranial pressure change, the duration was prolonged. No significant differences were found between the two groups in age, gender and Glasgow Coma Scale upon admission. Further, no differences were found in terms of the classification of computed tomography findings or the occurrence of pupillary abnormalities on admission. The patients in this study had not sustained either abdominal or thoracic trauma. Before inducing hypothermia, IL-6 levels in the arterial and internal jugular venous blood exceeded the normal range. Specifically, the internal jugular plasma levels were significantly higher than those in the arterial plasma. While IL-6 levels in the normothermic group did not decrease even at 4 days postinjury, the plasma cytokine levels fell at both sites sharply after moderate hypothermia. The cytokine suppression found in the hypothermic group continued even after rewarming in these patients showing an improved clinical course, but not in those whose condition worsened. In addition to these changes in cytokine levels, the Glasgow Outcome Scale at 6 months postinjury was significantly higher in the hypothermic group than in the normothermia group. Based on the above, this clinical study with its small patient sample size suggests the need for further prospective randomized studies to examine the role of cytokine suppression in the beneficial effects of moderate hypothermia in patients with traumatic brain injury.