Objective B-type natriuretic peptide(BNP) have been used widely in the diagnosis and prognosis of cardiogenic diseases, but the association between BNP concentrations and non-cardiogenic critical illness is unknown. Our study aims at investigating the role of natriuretic peptide testing for predicting 28-day mortality of non-cardiogenic critically ill patients in emergency department (ED). Methods A total of 255 non-cardiogenic critically ill patients treated in the emergency department were prospectively enrolled and a sample of blood was obtained for BNP measurement from each patient. A variety of clinical and laboratory variables were recorded. After 28 days, the vital status of each patient was ascertained and the association between BNP values at presentation and mortality was assessed. Results 71 patients died during the 28-day observation period. Median BNP concentrations at presentation among decedents were significantly higher than those of survivors (326. 0 ng/L versus 50. 9 ng/L,P <0. 001). The optimal BNP cut point for predicting 28-day mortality was 114. 0 ng/L. In a multivariate analysis (Cox-regression), a BNP concentration greater than 114. 0 ng/L(RR 7. 268, 95% CI 3. 864-13. 672) and an Acute Physiology and Chronic Health Evaluation Ⅱ (APACHE Ⅱ) scores greater than 20 (RR 3. 330, 95% CI 1. 815-6. 109)were the independent predictors of the 28-day mortality. BNP concentration alone had an area under the receiver operating characteristic curve (AUC) of 0. 825 for predicting mortality. Conclusions BNP concentrations at presentation are strong predictor of 28-day mortality in patients with non-cardiogenic critically ill patients in ED, which is better than Rapid Acute Physiology Score (RAPS) , Rapid Emergency Medicine Score (REMS) and APACHE Ⅱ score.