Autopsies are the gold standard for diagnostic accuracy; however, no recent study has analyzed autopsies in heart failure (HF).We reviewed 1241 autopsies (January 2000-May 2005) and selected 232 patients with HF. Clinical and autopsy diagnoses were analyzed and discrepancies categorized according to their importance regarding therapy and prognosis.Mean age was 63.3 ± 15.9 years; 154 (66.4%) patients were male. The causes of death at autopsy were end-stage HF (40.9%), acute myocardial infarction (17.2%), infection (15.9), and pulmonary embolism 36 (15.5). Diagnostic discrepancies occurred in 191 (82.3%) cases; in 56 (24.1%), discrepancies were related to major diagnoses with potential influence on survival or treatment; pulmonary embolism was the cause of death for 24 (42.9%) of these patients. In 35 (15.1%), discrepancies were related to a major diagnosis with equivocal influence on survival or treatment; in 100 (43.1%), discrepancies did not influence survival or treatment. In multivariate analysis, age (OR: 1.03, 95% CI: 1.008-1.052, P = 0.007) and presence of diabetes mellitus (OR: 0.359, 95% CI: 0.168-0.767, P = 0.008) influenced the occurrence discrepancies.Diagnostic discrepancies with a potential impact on prognosis are frequent in HF. These findings warrant reconsideration in diagnostic and therapeutic practices with HF patients.