The purpose of this study was to test the effectiveness, in patients with known aortic or brachiocephalic arterial injury, of five previously published radiographic criteria for excluding aortography in patients with blunt chest trauma. These criteria were (1) normal findings on erect chest radiograph; (2) normal aortic arch and left subclavian artery; (3) normal aortic arch, descending aorta, aortopulmonary window, tracheal position, and left paraspinal interface; (4) normal right paratracheal stripe and nasogastric tube position, and (5) normal aortic arch and tracheal and nasogastric tube position. One or more of these criteria were met in 6% to 25% of patient with major thoracic arterial injury, depending on the criteria used. Interestingly, two (6%) patients had radiographs that showed no specific signs of mediastinal hemorrhage, which indicates that the chest radiograph is limited in its sensitivity to detect major thoracic arterial injury. Because of these results, we do not believe that attempts to limit aortography in patients with supine film evidence of mediastinal abnormality, based on the absence of certain signs of mediastinal hemorrhage, are warranted. Furthermore, an abnormal radiograph cannot be relied on as the sole criterion for aortography if the goal of care is to detect as close to 100% of vascular injuries as possible.