The C-reactive protein/albumin ratio (CAR), a novel inflammation-based index, has been proved useful in predicting outcome of various diseases. We designed this study to explore the prognostic value of CAR in patients with traumatic brain injury (TBI). We retrospectively included 151 patients diagnosed with TBI and collected related clinical and laboratory data. Univariate and multivariate logistic regression were conducted to find independent risk factors of mortality. Then, we incorporated CAR into prognostic model and drew receiver operating characteristic (ROC) curve of models. Finally, we compared the predictive value of different models by evaluating the area under the receiver operating characteristic curves (AUC). In this study, a total of 54 patients had poor survival outcome with mortality rate of 35.8 %. Results of multivariate analysis showed that GCS score in admission (OR 0.700, 95 %Cl 0.570-0.860, p=0.001), acute kidney injury (AKI) (OR 3.952, 95Cl 1.631-9.577, p=0.002) and CAR (OR 1.202, 95Cl 1.039-1.390, p=0.013) were independently associated with in-hospital mortality. The AUC value of predictive model composed of the above three factors was higher than GCS or CAR alone. CAR is an independent risk factor of mortality in patients with TBI. Incorporating CAR into predictive model could increase the value in predicting outcome of TBI patients.