This study analyzed the relevance of plasma brain natriuretic peptide (BNP) and echocardiography in predicting cardiovascular events in a large population >70 years old with heart failure (HF). Three hundred four outpatients with HF (51.6% men, mean age 78.6) underwent transthoracic echocardiography and plasma BNP testing shortly before hospital discharge. Echocardiography was intended to reveal systolic dysfunction (left ventricular [LV] ejection fraction [EF] <50%) or diastolic dysfunction (EF > or =50% and abnormalities of ventricular relaxation). During 6-month follow-up, all-cause death and readmission were assessed. One hundred seventeen patients had diastolic dysfunction with preserved systolic LV function, and 187 had systolic dysfunction. At 6-month clinical follow-up, 33 subjects (10.9%) had died, and 62 (20.4%) needed readmission for cardiac decompensation. In all patients, univariate logistic regression demonstrated significant correlations between age (r = 0.14, p = 0.01), plasma BNP (r = 0.36, p = 0.0001), the EF (r = 0.16, p = 0.003), urea nitrogen (r = 0.35, p = 0.0001), serum creatinine (r = 0.27, p = 0.0001), and New York Heart Association (NYHA) class (r = 0.35, p = 0.0001) and the occurrence of cardiovascular events. In patients with HF in NYHA class III or IV, a BNP cut-off level of 200 pg/ml identified different outcomes (BNP <200 pg/ml in 1 of 20 events vs BNP >200 pg/ml in 55 of 85 events, p = 0.0001). In patients with HF who were >70 years old, BNP, NYHA class, and renal function predicted adverse outcome. In patients with severe HF, BNP was better than NYHA class in predicting future events.