Dyspnea is a common symptom in patients admitted to the Emergency Department (ED), and discriminating between cardiogenic and non-cardiogenic dyspnea is often a clinical dilemma. The initial diagnostic work-up may be inaccurate in defining the etiology and the underlying pathophysiology. The aim of this study was to evaluate the diagnostic accuracy and reproducibility of pleural and lung ultrasound (PLUS), performed by emergency physicians at the time of a patient's initial evaluation in the ED, in identifying cardiac causes of acute dyspnea. Between February and July 2007, 56 patients presenting to the ED with acute dyspnea were prospectively enrolled in this study. In all patients, PLUS was performed by emergency physicians with the purpose of identifying the presence of diffuse alveolar-interstitial syndrome (AIS) or pleural effusion. All scans were later reviewed by two other emergency physicians, expert in PLUS and blinded to clinical parameters, who were the ultimate judges of positivity for diffuse AIS and pleural effusion. A random set of 80 recorded scannings were also reviewed by two inexperienced observers to assess inter-observer variability. The entire medical record was independently reviewed by two expert physicians (an emergency medicine physician and a cardiologist) blinded to the ultrasound (US) results, in order to determine whether, for each patient, dyspnea was due to heart failure, or not. Sensitivity, specificity, and positive/negative predictive values were obtained; likelihood ratio (LR) test was used. Cohen's kappa was used to assess inter-observer agreement. The presence of diffuse AIS was highly predictive for cardiogenic dyspnea (sensitivity 93.6%, specificity 84%, positive predictive value 87.9%, negative predictive value 91.3%). On the contrary, US detection of pleural effusion was not helpful in the differential diagnosis (sensitivity 83.9%, specificity 52%, positive predictive value 68.4%, negative predictive value 72.2%). Finally, the coexistence of diffuse AIS and pleural effusion is less accurate than diffuse AIS alone for cardiogenic dyspnea (sensitivity 81.5%, specificity 82.8%, positive predictive value 81.5%, negative predictive value 82.8%). The positive LR was 5.8 for AIS [95% confidence interval (CI) 4.8-7.1] and 1.7 (95% CI 1.2-2.6) for pleural effusion, negative LR resulted 0.1 (95% CI 0.0-0.4) for AIS and 0.3 (95% CI 0.1-0.8) for pleural effusion. Agreement between experienced and inexperienced operators was 92.2% (p < 0.01) and 95% (p < 0.01) for diagnosis of AIS and pleural effusion, respectively. In early evaluation of patients presenting to the ED with dyspnea, PLUS, performed with the purpose of identifying diffuse AIS, may represent an accurate and reproducible bedside tool in discriminating between cardiogenic and non-cardiogenic dyspnea. On the contrary, US detection of pleural effusions does not allow reliable discrimination between different causes of acute dyspnea in unselected ED patients.