Objective To investigate the causes of death and evaluation of injury severity in patients with thoracic trauma so as to enhance the diagnosis and treatment of thoracic trauma. Methods A retrospective study was carried out in 687 patients with thoracic trauma, which were divided into different groups according to their condition of injury (chest injury group and multiple injuries group) and outcome (survival group and death group) and penetrating into pleural cavity (penetrating injury group and blunt trauma group), then trauma scores(revised trauma score,abbreviated injury scale,injury severity score,probability of survival)were compared respectively. In addition, the highrisk causes of trauma death were analyzed. Results Among 687 cases, there are 488 cases with blunt trauma and 199 cases with penetrating injury. The causes of trauma death in blunt trauma group were brain injury (10 cases) and acute respiratory failure (6 cases) and multiple organ dysfunction syndrome (4 cases) and hypovolemic shock (1 case). The causes of trauma death in penetrating trauma group were hypovolemic shock (9 cases). There were statistically difference of trauma score in the death group and the survival group（GCS：t=4.648,P=0.000; RTS:t=4.382,P=0.000;thoracic AIS：t=2.296,P=0.027;ISS：t=4.871，P=0.000; Ps：t=4.254,P=0.000）. There was no statistically difference of thoracic AIS in the chest injury group and the multiple injuries group (t=0.723, P=34.567), and there were statistical significances in RTS（t=2.553,P=0.032）, ISS（t=10.776,P=0.000）, Ps（t=3.868,P=0.007）. There were statistically difference of RTS（t=3.161,P=0.007）and ISS （t=4.118,P=0.005） in the blunt trauma survival group and penetrating injury survival group, and there was no statistical significance in Ps（t=0.857,P=97.453）. The blunt trauma death group had statistical difference compared with penetrating injury death group in trauma score（GCS：t=4.016,P=0.001，RTS:t=3.168,P=0.006;thoracic AIS：t=2.303, P=0.043;ISS：t=4.218,P=0.002;Ps：t=4.624,P=0.001）. The mortality of trauma was gradually increased with the trauma scores. The mortality was 10.7% when whole ISS was 20.25. The mortality in penetrating injury group was higher than that in blunt trauma group with the same ISS between two groups. Conclusion Applying trauma score is conducive to the judgement of trauma severity and optimizing clinical treatment. The death causes in blunt trauma group were more complex than in penetrating injury group. Severe trauma and multiple injuries are the main death causes of thoracic trauma.