Surgical management and prognosis of traction injuries of the brachial plexus depend on the accurate diagnosis of root avulsion from the spinal cord. Myelography, computerized tomography (CT) myelography, and recently magnetic resonance (MR) imaging have become the main radiological methods for preoperative diagnosis of cervical root avulsions. Most of the previous studies on the accuracy of CT myelography and MR imaging studies have correlated the radiological findings with the extraspinal surgical findings at brachial plexus surgery. Surgical experience shows that in many cases extraspinal findings diverge from intradural determinations. Consequently, only correlation with the intradural surgical findings will allow assessment of the factual accuracy of CT myelography and MR imaging studies. In a prospective study, 135 cervical roots (C5-8) were evaluated by CT myelography and/or MR imaging and further explored intradurally via a hemilaminectomy. The accuracy of the preoperative CT myelography-based diagnosis in relation to the intraoperative findings was 85%. On the other hand, MR imaging demonstrated an accuracy of only 52%. The most common reasons for false-positive or false-negative findings were: 1) partial rootlet avulsion; 2) intradural fibrosis; and 3) dural cystic lesions. Computerized tomography myelography scans using 1- to 3-mm axial slices prove to be the most reliable method to evaluate preoperatively the presence of complete or partial root avulsion in traumatic brachial plexus injuries. Because extradural judgment of cervical root avulsion can be unreliable, accurate assessment of intraspinal root avulsion enormously simplifies the decision concerning the choice of donor nerves for transplantation and/or neurotization during brachial plexus surgery.