Background Despite positive associations with rejection, the clinical value of B-type natriuretic peptide (BNP) monitoring in heart transplant recipients has not been established. We sought to determine the predictive value of changes in serial BNP level for identifying patients with acute allograft rejection. Methods BNP, hemodynamics and biopsies were obtained for 205 transplant recipients who underwent a total of 4,007 endomyocardial biopsy procedures. Samples analyzed were collected ≥180 days post-transplant, without evidence of rejection on the immediately preceding biopsy. Using a repeated-measures multivariate model, we assessed the association of change in BNP with Grade ≥3A (2R) rejection. We also determined predictive values of various cut-off thresholds of change in serial BNP levels to predict Grade ≥3A rejection. Results There were 47 episodes of Grade ≥3A rejection among the 1,350 samples analyzed. Median change in serial BNP (ΔBNP) for those with Grade ≥3A rejection was 20 pg/ml (IQR −26 to 169 pg/ml) and among those with Grade <3A rejection was −4 pg/ml (IQR −34 to 22 pg/ml, p = 0.003). On multivariate analysis, ΔBNP remained the most potent independent predictor of Grade ≥3A rejection (p = 0.001). ΔBNP >100 pg/ml predicted increased risk of Grade ≥3A rejection (OR = 5.3, p < 0.001) with high specificity (93.3%) and positive predictive value (13.0%) and excellent negative predictive value (97.3%). Conclusions Change in serial BNP level is an independent predictor of cardiac allograft rejection. With wide availability, rapid turnaround, low cost, favorable positive predictive value and excellent negative predictive value, serial BNP monitoring has several advantages for non-invasive monitoring of heart transplant recipients for acute cardiac allograft rejection.