Abstract Background/Introduction Acute heart failure (AHF) is one of the major causes of mortality, and identifying the patients at high risk of mortality at the time of admission is crucial to improve clinical outcomes. Although some risk prediction models for patients with AHF have been proposed mainly from randomized clinical trials, the patients in such studies tend not to be similar to those in the real world. Recently, BIOSTAT risk score and AHEAD score derived from two large-scale registry dataset are proposed as useful risk stratification tools for patients with AHF. However, these scores have not been well externally validated and their prognostic prediction performance has not been directly compared. Purpose To validate and compare prognostication of BIOSTAT risk score and AHEAD score in AHF patients. Methods Patients who consecutively admitted to the cardiac intensive-care unit in our institution with a diagnosis of AHF from 2007 to 2011 were analyzed. Among them, patients with acute coronary syndrome, dialysis, malignancy were excluded. BIOSTAT risk score was calculated using 5 factors (age, blood urea nitrogen, BNP, hemoglobin, prescription of beta blockers), and AHEAD score was also calculated with 5 factors (atrial fibrillation, hemoglobin, age, creatinine, and diabetes mellitus). We also developed AHEAD + BNP model incorporating BNP into AHEAD score. Endpoint was 1-year all-cause death. Results Overall, 591 eligible patients were enrolled (mean age was 70±14 years old, 64.8% were male) and 96 patients (16.2%) died during the follow-up of 1-year. The median [interquartile range] of AHEAD score and BIOSTAT risk score were 2 [1–3] and 3 [2–4], respectively. The areas under the curves of receiver operating characteristic curve (AUC) were 0.66 for AHEAD, 0.68 for AHEAD + BNP, and 0.72 for BIOSTAT, respectively. The calibration plots for AHEAD, AHEAD + BNP, and BIOSTAT models showed good calibration (Hosmer-Lemeshow test: p=0.89, 0.74, and 0.74, respectively). The BIOSTAT model's AUC was significantly higher compared to AHEAD (p=0.018) and marginally statistically higher compared to AHEAD + BNP (p=0.054). However, BIOSTAT model showed statistically significant net reclassification improvement compared to both AHEAD (NRI: 0.43, p<0.001) and AHEAD + BNP (NRI: 0.43, p<0.001). Conclusion The BIOSTAT score comprised of five readily available clinical variables predict 1-year mortality of patients with AHF with good discrimination and calibration. ROC curves Funding Acknowledgement Type of funding source: None