Pericardial effusion is associated with an abnormal increase in respiratory variation in mitral flow velocity. However, the relation of the changes in flow velocity to pericardial pressure, hemodynamics and two-dimensional echocardiographic findings is not established. Therefore, 11 sedated dogs with extensive hemodynamic instrumentation were studied with two-dimensional and Doppler echocardiography during four stages of progressively larger pericardial effusion. During all stages of effusion, respiratory variation in peak mitral flow velocity in early diastole and left ventricular isovolumetric relaxation time was increased compared with baseline (p less than 0.05). This increase was seen at the earliest stage of effusion (mean pericardial pressure 4.2 +/- 1.4 versus -0.8 +/- 0.9 mm Hg at baseline, p less than 0.05), and preceded the appearance of unequivocal diastolic right heart collapse in every dog. Maximal respiratory variation coincided with the appearance of right atrial collapse (mean pericardial pressure 7.1 +/- 2.4 mm Hg; mean inspiratory decrease in aortic pressure 9.5 +/- 2.6 mm Hg; mean aortic pressure 88.2 +/- 15.2 versus 102.2 +/- 11.2 mm Hg at baseline, p less than 0.05; and cardiac output 3.8 +/- 1.2 versus 5.5 +/- 1.3 liters/min at baseline, p less than 0.05), but did not increase at stages associated with more severe hemodynamic compromise. In addition, the respiratory changes in peak mitral flow velocity in early diastole were associated with simultaneous changes in the diastolic transmitral pressure gradient. It is concluded that in this model of acute pericardial effusion 1) increased respiratory variation in early diastolic mitral flow velocity, peak mitral flow velocity in early diastole and left ventricular isovolumetric relaxation time occurs almost immediately as pericardial pressure increases and persists at all stages of increasing pericardial effusion; 2) the abnormal respiratory variation occurs before equalization of intracardiac pressures and before the onset of unequivocal right heart collapse; 3) the respiratory variation occurs as a result of changes in the diastolic transmitral pressure gradient; and 4) the magnitude of the respiratory change is not necessarily predictive of pericardial pressure or severity of hemodynamic compromise, especially at the more severe stages of pericardial effusion.