Myocardial and cerebral blood flow can be generated during cardiac arrest by techniques that manipulate intrathoracic pressure. Augmentation of intrathoracic pressure by high-pressure ventilation simultaneous with compression of the chest in dogs has been shown to produce higher flows to the heart and brain, but has limited usefulness because of the requirement for endotracheal intubation and complex devices. A system was developed that can produce high intrathoracic pressure without simultaneous ventilation by use of a pneumatically cycled vest placed around the thorax (vest cardiopulmonary resuscitation [CPR]). The system was first tested in a short-term study of the maximum achievable flows during arrest. Peak vest pressures up to 380 mm Hg were used on eight 21 to 30 kg dogs after induction of ventricular fibrillation and administration of epinephrine. Microsphere-determined myocardial blood flow was 108 $ 17 ml/min/100 g (100 $ 16% of prearrest flow) and cerebral flow was 51 $ 12 ml/min/100 g (165 $ 39% of prearrest). Severe lung or liver trauma was noted in three of eight dogs. If peak vest pressure was limited to 280 mm Hg, however, severe trauma was no longer observed. A study of the hemodynamics during and survival from prolonged resuscitation was then performed on three groups of seven dogs. Vest CPR was compared with manual CPR with either conventional (300 newtons) or high (430 newtons) sternal force. After induction of ventricular fibrillation, each technique was performed for 26 min. Defibrillation was then performed. After 20 min of resuscitation, vest CPR produced a myocardial flow of 54 $ 13 ml/min/100 g (40 $ 9% of prearrest flow) and a cerebral flow of 37 $ 4 ml/min/100 g (99 $ 11% of prearrest). With conventional sternal force, manual CPR produced lower myocardial and cerebral flows than did the vest method (p< .04), and resulted in fewer next-day survivors (7/7 for vest vs 1/7 for manual,p< .003). With high sternal force, flows were similar to those obtained with the vest, but more dogs had severe rib or liver trauma (0/7 for vest vs 4/7 for manual,p< .04), and there were still fewer survivors than with the vest method (3/7,p< .04 vs vest). Thus, at very high pressures, vest CPR can generate essentially normal myocardial and cerebral flow, but can also produce severe trauma. At lower pressures, vest CPR can improve survival after cardiac arrest, while producing less trauma than manual CPR performed with sufficient compression to generate comparable flows. Vest CPR warrants study in man as a potential means for augmenting flow during cardiac arrest without the need for endotracheal intubation and simultaneous ventilation.