PURPOSE: Patients with traumatic brain injury (TBI) were referred from other hospitals for further management. In addition, patients routinely underwent computed tomography examinations of the head (HCT) in the referral hospitals. The purpose of this study was to evaluate retrospectively the utility of routine HCT scans according to the severity of TBI. METHODS: Patients with TBI referred to our hospital between December 2005 and July 2008 were included in this study. We investigated HCT findings, indications for repeat HCT examinations (routine versus a neurological change), and neurosurgical interventions. The head injury severity was divided into three categories according to the Glasgow Coma Scale (GCS) score, including mild, moderate, and severe TBI. The use of neurosurgical interventions between patients who underwent routine HCT scans and patients who underwent HCT scans for a neurological change were compared according to the severity of TBI. RESULTS: A total of 81 patients met the entry criteria for this study. Among these patients, 67%(n=54) of the patients underwent HCT scans on a routine basis, whereas 33%(n=27) of the patients underwent HCT scans for a neurological change. A total of 21 patients showed signs of a worsening condition on the HCT scans. Neurosurgical intervention was required for 23(28.4%) patients. For patients who underwent routine HCT examinations, no patient with mild TBI underwent a neurosurgical intervention. However, one patient with moderate TBI and three(13%) patients with severe TBI underwent neurosurgical interventions. The kappa index, the level of agreement for HCT indications of intervention and referral reasons for intervention, was 0.65 for high hierarchy hospitals and 0.06 for low hierarchy hospitals. CONCLUSION: Routine serial HCT examinations in the referred hospitals would be useful for patients with severe head injury and for patients from low hierarchy hospitals where no emergency physicians or neurosurgeons are available.