OBJECTIVE Animal studies have shown an aortic-carotid artery pressure difference during cardiopulmonary resuscitation (CPR), which compromises cerebral perfusion. This pressure difference is most marked with prolonged CPR and can be abolished with administration of high doses of epinephrine. To better understand the mechanism of cerebral blood flow during CPR in humans, we determined the aortic-carotid artery pressure difference, the cephalic perfusion pressure (the carotid artery-jugular vein pressure difference), and thoracic inlet venous "valving" (the central venous-jugular vein pressure difference), while administering standard doses of epinephrine. DESIGN Prospective study with randomization as to which side the carotid artery was catheterized. SETTING The resuscitation room of a large urban hospital's emergency department. PATIENTS Fifteen adults in normothermic, nontraumatic prehospital cardiac arrest treated according to Advanced Cardiac Life Support guidelines, including administration of 1 mg epinephrine iv every 5 mins. INTERVENTIONS The descending aorta, cervical common carotid artery, internal jugular vein, and central venous system were catheterized. Pressures were recorded during standard CPR for 5 mins after administration of 1 mg epinephrine iv. MEASUREMENTS AND MAIN RESULTS Most patients received CPR for greater than 20 mins before the first epinephrine dose and for greater than 45 mins before pressure recording as described above. There was no significant difference between aortic and carotid artery compression and relaxation phase pressures. The mean +/- SD compression central venous-jugular vein pressure difference was 22.1 +/- 15.0 mm Hg, and the mean cephalic perfusion pressure was 20.8 +/- 19.5 mm Hg. CONCLUSIONS There is no clinically important aortic-carotid artery pressure difference during human CPR using the standard dose of epinephrine, even with prolonged CPR. Despite carotid artery patency and thoracic inlet venous valving, the cephalic perfusion pressure is low during CPR in humans.