SUMMARY Cardiopulmonary resuscitation (CPR) has been thought to produce blood flow by compression of the heart between the sternum and spine, termed "external cardiac massage," but there has been-no direct experimental documentation of this proposed mechanism.Micromanometric pressure recordings were synchronized with cineangiograms during mechanical CPR in 17 dogs with induced ventricular fibrillation. Chest compression produced equivalent pressure increases in the aorta (Ao) and right atrium (RA) (Ao 32 + 14 mm Hg, RA 30 ± 14 mm Hg; NS), a linear relationship between aortic and intrapleural pressures (r = 0.87, p < 0.001) over a wide range of induced pressures, cineangiographic blood flow through both left-heart chambers, and a pressure gradient (21 ± 14 mm Hg) between all intrathoracic cardiovascular compartments and the jugular veins that resulted from closure of venous valves at the thoracic inlets. Simultaneous chest compression and lung inflation significantly increased all intrathoracic vascular pressures, the aortojugular venous gradient (42 ± 13 mm Hg, p < 0.05 vs chest compression alone), electromagnetically determined carotid arterial blood flow (1.75 ± 0.81 ml/min/kg vs 0.51 ± 0.27 mI/min/kg during chest compression alone, p < 0.005), and angiographic left-heart flow.We conclude that blood flow during CPR results principally from an increased intrathoracic pressure and that there is selective flow to the brachiocephalic vascular bed because of the arteriovepous gradient produced by venous valves at the thoracic inlets. Greater intrathoracic pressure resulting from simultaneous inflation and compression improves left-heart flow. The left heart is therefore a conduit, not a pump, during CPR.THE ABILITY of closed-chest cardiopulmonary resuscitation (CPR) to produce an arterial pulse and perfuse vital organs in the arrested circulation was demonstrated experimentally and clinically in 1960 and termed "external cardiac massage."' As the name implied, the technique was thought to produce selective ventricular compression between the sternum and spine, propelling blood forward through the systemic vasculature. The technique was quickly accepted and its efficacy documented in clinical reports.2However, early hemodynamic observations during experimental CPR by Weale and Rothwell-Jackson' showed that chest compression resulted in an equal increase in arterial and venous pressures recorded in the iliac vessels, and the absence of a significant pressure gradient was believed to argue against antegrade flow. right atrium and arterial bed could result in retrograde flow or reflux into the extrathoracic venous bed. Rudikoff and co-workers" showed the absence of a significant net aortic-central venous pressure gradient during chest compression and documented that the pressures in all cardiac chambers and in the thoracic aorta increase proportionately with the increase in intrapleural pressure. These findings suggest that vascular pressures and flow recorded during chest compression are dependent upon the generated i...