Pulmonary artery pressure was noninvasively estimated by three Doppler echocardiographic methods in 50 consecutive patients undergoing cardiac catheterization. First, a systolic transtricuspid gradient was calculated from Doppler-detected tricuspid regurgitation; clinical jugular venous pressure or a fixed value of 14 mm Hg was added to yield systolic pulmonary artery pressure. Second, acceleration time from pulmonary flow analysis was used in a regression equation to derive mean pulmonary artery pressure. Third, right ventricular isovolumic relaxation time was calculated from Doppler-determined pulmonary valve closure and tricuspid valve opening; systolic pulmonary artery pressure was then derived from a nomogram. In 48 patients (96%) at least one of the methods could be employed. A tricuspid pressure gradient, obtained in 36 patients (72%), provided reliable prediction of systolic pulmonary artery pressure. The prediction was superior when 14 mm Hg rather than estimated jugular venous pressure was used to account for right atrial pressure. In 44 patients (88%), pulmonary flow was analyzed. Prediction of mean pulmonary artery pressure was unsatisfactory (r = 0.65) but improved (r = 0.85) when only patients with a heart rate between 60 and 100 beats/min were considered. The effect of correcting pulmonary flow indexes for heart rate was examined by correlating different flow indexes before and after correction for heart rate. There was a good correlation between corrected acceleration time and either systolic (r = -0.85) or mean (r = -0.83) pulmonary artery pressure. Because of a high incidence of arrhythmia, right ventricular relaxation time could be determined in only 11 patients (22%). Noninvasive prediction of pulmonary artery pressure is feasible in most patients.(ABSTRACT TRUNCATED AT 250 WORDS)