The accuracy of the logistic EuroSCORE (logES), a widely used risk prediction algorithm for cardiac surgery including aortic valve surgery, usually overestimates observed perioperative mortality. Elevated brain natriuretic peptide (BNP) in symptomatic patients with aortic stenosis (AS) is associated with a poor short-term outcome after aortic valve replacement. We aimed to compare BNP with the logES for predicting short- and long-term outcome in symptomatic patients with severe AS undergoing aortic valve replacement. We prospectively studied 144 consecutive patients referred for aortic valve replacement (42% women, 73 +/- 9 years, mean aortic gradient 51 +/- 18 mm Hg, and left ventricular ejection fraction 61 +/- 11%) undergoing either isolated aortic valve replacement (58%) or combined to bypass grafting. Both plasma BNP and logES was estimated before surgery. The median BNP plasma level and logES were 157 pg/ml (interquartile range [IQR] 61 to 440) and 6.6% (IQR 4.2 to 12.2), respectively. The perioperative mortality was 6% and the overall mortality by the end of the study was 13%. Patients with logES >10.1% (upper tertile) had a higher risk of dying over time (hazard ratio [HR] 2.86, p = 0.037), as had patients with BNP >312 pg/ml (HR 9.01, p <0.001). Discrimination (based on C statistic) and model performance (based on Akaike information criterion) were better for BNP than for logES. At the bivariable analysis, only BNP was an independent predictor of death (HR 8.2, p = 0.002). Preoperative BNP was even more accurate than logES in predicting outcome. In conclusion, in symptomatic patients with severe AS, high preoperative BNP plasma level and high logES confirm their predicting value for short- and long-term outcome.