Traumatic rupture of the thoracic aorta is a justifiably feared condition. Some authors have proposed the use of computed tomography as a non-invasive means of diagnosis. We report two cases where computed tomographic scans were misinterpreted, leading to erroneous diagnoses and inappropriate referrals. The pitfalls of using a cross-sectional imaging technique to diagnose a transverse lesion and the relative ease and accuracy of aortography are discussed.