OBJECTIVE: To evaluate the diagnostic performance of transthoracic sonography of the lung in view of an acute suspicion of pulmonary embolism. METHODS: The diagnosis of pulmonary embolism was made within 48 hours of admission with the use of a validated strategy that combined a lung scan, lower limb sonography, helical computed tomography, and pulmonary angiography in 74 consecutive patients. All patients had a 3-month follow-up. Transthoracic sonography of the lung and pleura was performed in a blinded fashion by investigators unaware of the diagnostic process. It was interpreted as suggestive or not suggestive of pulmonary embolism on the basis of previously published descriptions. The results were kept apart from the patient files until the end of the study. RESULTS: Pulmonary embolism was confirmed in 31 (42%) of the 74 patients. None of the patients classified as pulmonary embolism-negative had a thromboembolic event during the 3-month follow-up. The diagnostic performance of transthoracic sonography was as follows: sensitivity, 71% (95% confidence interval, 52%-85%); specificity, 77% (95% confidence interval, 61%-88%); positive predictive value, 69% (95% confidence interval, 50%-83%); and negative predictive value, 79% (95% confidence interval, 63%-89%). The finding of an isolated pleural infarction was significantly more frequent in pulmonary embolism-positive patients compared with pulmonary embolism-negative patients (P = .0001). CONCLUSIONS: The results from this study suggest that transthoracic sonography of the lung does not fulfill satisfactory performance criteria in view of a clinical suspicion of pulmonary embolism.