Circulating microRNAs (miRNAs) present in the serum/plasma are characteristically altered in many pathological conditions, and have been employed as diagnostic markers for specific diseases. We examined if plasma miRNA levels are altered in patients with traumatic brain injury (TBI) relative to matched healthy volunteers, and explored their potential for use as diagnostic TBI biomarkers. The plasma miRNA profiles from severe TBI patients (Glasgow Coma Scale [GCS] score ≤8) and age-, gender-, and race-matched healthy volunteers were compared by microarray analysis. Of the 108 miRNAs identified in healthy volunteer plasma, 52 were altered after severe TBI, including 33 with decreased and 19 with increased relative abundance. An additional 8 miRNAs were detected only in the TBI plasma. We used quantitative RT-PCR to determine if plasma miRNAs could identify TBI patients within the first 24 h post-injury. Receiver operating characteristic curve analysis indicated that miR-16, miR-92a, and miR-765 were good markers of severe TBI (0.89, 0.82, and 0.86 AUC values, respectively). Multiple logistic regression analysis revealed that combining these miRNAs markedly increased diagnostic accuracy (100% specificity and 100% sensitivity), compared to either healthy volunteers or orthopedic injury patients. In mild TBI patients (GCS score > 12), miR-765 levels were unchanged, while the plasma levels of miR-92a and miR-16 were significantly increased within the first 24 h of injury compared to healthy volunteers, and had AUC values of 0.78 and 0.82, respectively. Our results demonstrate that circulating miRNA levels are altered after TBI, providing a rich new source of potential molecular biomarkers. Plasma-derived miRNA biomarkers, used in combination with established clinical practices such as imaging, neurocognitive, and motor examinations, have the potential to improve TBI patient classification and possibly management.