Background Heart failure complicating acute myocardial infarction marks an ominous prognosis. Killip and Kimball's classification of heart failure remains a useful tool in these patients. Lung ultrasound can detect pulmonary congestion but its usefulness in this scenario is unknown. Objective To investigate the diagnostic accuracy of lung ultrasound to predict heart failure in patients with acute myocardial infarction. Methods Patients admitted with acute myocardial infarction and without heart failure were evaluated with a lung ultrasound. The presence of B-lines was recorded and counted. The presence of new heart failure (Killip Class B, C, or D) during hospitalization was evaluated by a cardiologist blinded to the results of lung ultrasound. A ROC curve analysis was done to evaluate the diagnostic accuracy of B-lines to predict heart failure. Results 200 patients were included. Three patients were diagnosed with cardiogenic shock, 5 with acute pulmonary edema, and 17 with mild heart failure. Patients who develop heart failure had a median of 14 B-lines, however, patients who remained in Killip class A had a median of 2 ( p = 0,0001). The area under the ROC curve of the sum of B-lines to predict any form of heart failure was 0,91 (CI95% 86–97). The best cut-off value was 5 B-lines, with a sensitivity of 88% (IC95% 68,8–97,5) and specificity of 81% (IC95% 73,9–86,2). Conclusion Lung ultrasound done at admission can help to predict heart failure In patients with acute myocardial infarction.