Introduction. Traumatic brain injuries (TBIs) are very common in paediatric populations, in which they are also a leading cause of death. Computed tomography (CT) overuse in these populations results in ionization radiation exposure, which can lead to lethal malignancies. The aims of this study were to investigate the accuracy of serum S100B levels with respect to the detection of cranial injury in children with mild TBI and to determine whether decisions regarding the performance of CT can be made based on biomarker levels alone. Materials and Methods. This was a single-center prospective cohort study that was carried out from December 2016 to December 2017. A total of 80 children with mild TBI who met the inclusion criteria were included in the study. The patients were between 2 and 16 years of age. We determined S100B protein levels and performed head CTs in all the patients. Results. Patients with cranial injury, as detected by CT, had higher S100B protein levels than those without cranial injury (p<0.0001). We found that patients with cranial injury (head CT+) had higher mean S100B protein levels (0.527 μg L−1, 95% confidence interval (CI) 0.447–0.607 μg L−1) than did patients without cranial injury (head CT−) (0.145 μg L−1, 95% CI 0.138–0.152 μg L−1). Receiver operating characteristic (ROC) curve analysis clearly showed that S100B protein levels differed between patients with and without cranial injury at 3 hours after TBI (AUC = 0.893, 95% CI 0.786–0.987, p=0.0001). Conclusion. Serum S100B levels cannot replace clinical examinations or CT as tools for identifying paediatric patients with mild head injury; however, serum S100B levels can be used to identify low-risk patients to prevent such patients from being exposed to radiation unnecessarily.