Emergency thoracotomy is a standard procedure in the management of cardiac arrest in patients sustaining severe trauma. We examined the records of 463 moribund trauma patients treated at our institution from 1980 to 1990 to refine indications for emergency thoracotomy. Patients underwent thoracotomy either in the emergency department (ED) (n = 424) or in the operating room (OR) (n = 39) as a component of continuing resuscitation after hospital arrival. The survival rate was 13% (61 of 463) overall, 2% (3 of 193) for blunt, 22% (58 of 269) for all penetrating, 8% (10 of 131) for gunshot, 34% (48 of 141) for stab-wound patients, and 54% (21 of 39) for patients who underwent emergency thoracotomy in the OR. Survival correlated with the physiologic status of patients both on initial evaluation in the field by paramedics and on arrival at the ED. Patients with penetrating trauma and in profound shock (BP < 60 mm Hg) or mild shock (BP 60–90 mm Hg) with subsequent cardiac arrest had survival rates of 64% (27 of 42) and 56% (30 of 54), respectively. None of the patients with absent signs of life, defined as full cardiopulmonary arrest with absent reflexes (n = 215), on initial assessment by paramedics in the field, survived. We conclude that (1) no emergency thoracotomy should be performed if no signs of life are present on the initial prehospital field assessment; (2) emergency thoracotomy is an indicated procedure in most patients sustaining penetrating trauma; (3) blunt traumatic cardiac arrest is a relative contraindication to emergency thoracotomy.