Nearly 1/3 of patients with heart failure (HF) fail to respond to cardiac resynchronization therapy (CRT). The purpose of this study was to evaluate the value of preimplantation brain natriuretic peptide (BNP) in predicting the clinical response to CRT. We retrospectively analyzed 164 patients who underwent CRT. Patients with New York Heart Association functional class III or IV HF symptoms despite maximal medical therapy, who were not on inotropic medications, had left ventricular ejection fraction < or =35%, and QRS duration >130 ms were included in the study. CRT response in patients who survived at 6-month follow-up was defined as no HF hospitalization and improvement of > or =1 grades in the New York Heart Association classification. BNP assays were performed before implantation and at 6-month follow-up. Patients had ischemic (47%) or nonischemic (53%) cardiopathy. Responders (n = 107) and nonresponders (n = 57) had similar baseline characteristics. Cardiac death and hospitalization for HF occurred in 5 (4.7%) and 18 (31.6%) patients, respectively. CRT responders compared with nonresponders exhibited higher preimplantation BNP levels (800 +/- 823 vs 335 +/- 348 pg/ml, p = 0.0002) and a significant reduction in the QRS duration after implantation (-6 +/- 34 vs +7 +/- 32 ms, p = 0.048). The preimplantation BNP was the only independent predictor of the CRT response (p = 0.001). A BNP value > or =447 pg/ml demonstrated a sensitivity of 62% and specificity of 79% in identifying CRT response. In a subgroup of 41 patients who underwent Doppler tissue imaging analysis, the preimplantation BNP was higher in patients presenting with intraventricular dyssynchrony (845 +/- 779 vs 248 +/- 290 pg/ml, p = 0.04). In conclusion, the preimplantation BNP value independently predicts CRT response and was superior to QRS duration reduction in identifying CRT responders.