Transthoracic echocardiography and continuous wave Doppler were prospectively performed in 132 out-patients with suspicion of pulmonary embolism, and who had no previous history of severe cardiac or pulmonary disease. Bedside echocardiography determined diagnosis other than pulmonary embolism in 55 patients. Further study was completed in 70 patients; pulmonary embolism was found in 31 and excluded in 39. Significant differences were found as regards right ventricular diameter (27 +/- 8 vs 22 +/- 5 mm, P < 0.001), left ventricular diameter (41 +/- 9 vs 49 +/- 7 mm, P < 0.001), right over left ventricular diameter ratio (0.67 +/- 0.23 vs 0.43 +/- 0.15, P < 0.0001), tricuspid regurgitant flow peak velocity (2.9 +/- 0.4 vs 2.4 +/- 0.7 m.s-1, P < 0.0001), and abnormal septum motion (12 vs 4, P < 0.01). Multivariate analysis of echocardiographic data included a tricuspid regurgitant flow peak velocity greater than 2.5 m.s-1 and a right over left ventricular diameter ratio greater than 0.5 in a logistic model (sensitivity 93%, specificity 81%). The combination of echocardiographic and non-echocardiographic data included the two previous echocardiographic variables, together with signs of deep vein thrombosis, a deep S wave in lead D1, and a Q wave in lead D3 on the electrocardiogram in a logistic model (sensitivity 96%, specificity 83%). It can be concluded that emergency echocardiography, alone or combined with clinical examination and electrocardiogram, satisfactorily predicts acute pulmonary embolism.