B-type natriuretic peptide (BNP) has emerged as an important marker of ventricular wall stress and is predictive of hemodynamic abnormalities in heart transplantation despite "preserved" systolic function. We evaluated the capacity of BNP to predict deaths due to allograft failure in 62 patients long after heart transplantation (mean 5 +/- 2.5 years). Based on the median tendency of measurement of BNP in the absence of rejection during stable surveillance, 2 distinct patient groups were identified as having low BNP (n = 39, < 250 pg/ml; median BNP 70 pg/ml) and high BNP (n = 23, > or =250 pg/ml; median BNP 592 pg/ml). No differences between the 2 BNP groups were noted with regard to age, gender, race, time after transplantation, diabetes mellitus, hypertension, and hyperlipidemia with measurement of BNP. Multivariable analysis showed that decreased left ventricular ejection fraction, angiographic coronary artery disease, and increased serum creatinine were independent predictors of elevated BNP. Cardiac deaths were significantly greater in those with high BNP levels (35%) than in those with low BNP (2.5%, p = 0.01). Absence of significant angiographic coronary artery disease coupled with a BNP of < 250 pg/ml was associated with the lowest event rate (0%), whereas patients with coronary artery disease and BNP > or =250 pg/ml exhibited a 50% cardiac death rate (p <0.01 for trend). Cox's model confirmed that increased BNP and decreased left ventricular ejection fraction are independent predictors of poor survival. Survival analysis associated lower BNP levels with an excellent long-term survival rate (95%) and higher BNP levels with a markedly decreased survival rate (60%, p = 0.002). Higher BNP levels in patients long after heart transplantation are associated with allograft dysfunction and cardiac allograft vasculopathy and are strongly and independently predictive of cardiovascular death.