We directly compared the transmitral inflow pattern during preload reduction and pulmonary venous flow velocities to determine left ventricular end-diastolic pressure (LVEDP) in 78 patients who underwent left heart catheterization. Transmitral inflow indexes (A-wave duration, ratio of peak flow velocity of early diastole [E] to peak flow velocity of late diastole during atrial contraction [A] [E/A ratio]) at rest and during the Valsalva maneuver (30 mm Hg for 15 seconds) and indexes of pulmonary venous flow (velocity and duration of the atrial reversal) were obtained. Fair correlations existed between LVEDP (mean 15+/-6 mm Hg) and the percentage decrease in the E/A ratio (r = 0.72), increase in duration of A wave during the Valsalva maneuver (r = 0.60), flow velocity of atrial reversal (r = 0.58), and difference of duration of atrial flow reversal and A wave (r = 0.62) (all P<.001). While sensitivity, specificity, and diagnostic accuracy to detect an elevated LVEDP were comparable, technically adequate Doppler recordings were obtained more often for the mitral inflow during the Valsalva maneuver than for the pulmonary venous flow (72 versus 66 patients, P< 0.05).