N-terminal pro-brain-type natriuretic peptide (NT-proBNP) is currently used for risk stratification in acute pulmonary embolism (PE). We aimed to clarify the impact of renal function on the validity of the NT-proBNP based prognosis, assuming that the biomarker is accumulated in renal insufficiency. The NT-proBNP based prediction of PE related in-hospital death was investigated according to renal function in 329 patients with acute PE. The normalized NT-proBNP ratios (NT-proBNP level divided by the age-adjusted normal upper range) were inversely correlated (r = -0.414, P < 0.001) to the estimated glomerular filtration rates (eGFR). A cut-off point of ≥ 2.5 for the normalized NT-proBNP ratio was found to be best for the prediction of mortality (AUC 0.716, 95% CI 0.626-0.805, P < 0.001) and was a significant predictor for death in univariate and multivariate analysis. A normalized NT-proBNP ratio ≥ 2.5 was a significant predictor for PE-mortality only in patients with an eGFR ≥ 60 ml/min/1.73 m². Renal insufficiency significantly predicted mortality in univariate but not in multivariate analysis. High-risk PE and cerebrovascular diseases were significantly more frequent in renal dysfunction and significantly predicted death in univariate and multivariate analysis. The validity of the NT-proBNP based short-term prognosis might be limited in renal dysfunction not only by accumulation, but also because renal insufficiency itself and concurrent conditions are contributing to PE related mortality.