Aortic injuries during blunt thoracic impacts can lead to life threatening hemorrhagic shock and potential exsanguination. Experimental approaches designed to study the mechanism of aortic rupture such as the testing of cadavers is not only expensive and time consuming, but has also been relatively unsuccessful. The objective of this study was to develop a computer model and to use it to predict modes of loading that are most likely to produce aortic ruptures. Previously, a 3D finite element model of the human thorax was developed and validated against data obtained from lateral pendulum tests. The model included a detailed description of the heart, lungs, rib cage, sternum, spine, diaphragm, major blood vessels and intercostal muscles. However, the aorta was modeled as a hollow tube using shell elements with no fluid within, and its material properties were assumed to be linear and isotropic. In this study fluid elements representing blood have been incorporated into the model in order to simulate pressure changes inside the aorta due to impact. The current model was globally validated against experimental data published in the literature for both frontal and lateral pendulum impact tests. Simulations of the validated model for thoracic impacts from a number of directions indicate that the ligamentum arteriosum, subclavian artery, parietal pleura and pressure changes within the aorta are factors that could influence aortic rupture. The model suggests that a right-sided impact to the chest is potentially more hazardous with respect to aortic rupture than any other impact direction simulated in this study. The aortic isthmus was the most likely site of aortic rupture regardless of impact direction. The reader is cautioned that this model could only be validated on a global scale. Validation of the kinematics and dynamics of the aorta at the local level could not be done due to a lack of experimental data. It is hoped that this model will be used to design experiments that can reproduce field relevant aortic ruptures in the laboratory. Only after such experiments have been run, can local validation be examined and the model judged to be acceptable or unacceptable.