Mitral valve closure during cardiopulmonary resuscitation may result from direct cardiac compression. An alternative hypothesis is that with a rise in intrathoracic pressure, mitral valve closure can occur but may be influenced by whether the lungs are inflated or deflated. To test this hypothesis, we placed a large-bore cannula into the thoraces of 11 dogs. Intrathoracic pressure was changed by inflating and deflating the thorax through the cannula while the airway was open, as well as by inflating and deflating the lungs with the thoracic cannula clamped. Mitral valve motion was observed with two-dimensional echocardiography from the right chest wall or esophagus in eight of the dogs. With a rise in intrathoracic pressure from thoracic inflation, all eight dogs showed closure of the mitral valve, while with thoracic deflation, all showed mitral valve opening. With lung inflation and deflation alone, however, the mitral valve remained open throughout the cycle. In seven dogs, with thoracic inflation, the peak gradient from the left ventricle to the left atrium was (mean +/- SEM) 18 +/- 4 mm Hg and the average gradient was 7 +/- 3 mm Hg, while with lung inflation alone, the average gradient was -1 +/- 1 mm Hg (p less than 0.01 vs. thoracic inflation). Thus, mitral valve closure, with concomitant retrograde pressure gradients, can be produced by intrathoracic pressure changes with accompanying lung deflation. With lung inflation alone, however, the mitral valve remains open, and there are no significant transmitral pressure gradients. We conclude that intrathoracic pressure changes can cause the mitral valve to close or to remain open, depending on how intrathoracic pressure is generated.