Background: The purpose of this systematic review was to synthesize evidence based on existing studies on the ability of initial imaging to predict mortality in severe traumatic brain injuries (TBIs) in pediatric patients. Methods: An experienced librarian searched for all existing studies based on the inclusion and exclusion criteria. The studies were screened by two blinded reviewers. The data was extracted to calculate the sensitivity (SN), specificity (SP), positive predictive value (PPV), negative predicted value (NPV), area under the curve (AUC), and receiver operating characteristic (ROC) for extradural hematoma (EDH), subdural hematoma (SDH), traumatic subarachnoid hemorrhage (tSAH), skull fractures, and edema. Results: Of the 3277 studies included in the search, data could only be extracted from 22 studies. There were a total of 2219 patients, 747 females, and 1461 males. 564 patients died and 1651 survived. 293 patients had SDH, 76 had EDH, 347 had tSAH, 244 had skull fractures, and 416 had edema. Seven of the studies had sufficient data to calculate the AUC, ROC, and generate a forest plot for the imaging findings. Conclusions: Out of the different CT scan findings, brain edema had the highest SN, PPV, NPV, and AUC. EDH had the highest SP to predict in hospital mortality.