Abstract Background Body Mass Index (BMI) has been recognized as a prognosticator for heart failure (HF) in that higher BMI relates to lower mortality. However, limited data are available whether body composition underneath BMI would further differentiate patients with a better or worse outcome, especially in those with lower or normal BMI. Purpose To investigate the characteristics of body fat and body muscle in HF patient groups with different BMI, and their impact on prognosis. Methods Between July 2021 and July 2022, hospitalized HF patients who were suitable for the examination underwent body composition analysis by using bioelectrical analyses (BIA) in the stable state after decongestion therapy and before discharge. Percentage of body fat (PBF) and percentage of body muscle (PBM) were calculated by dividing the mass of body fat, or body muscle, respectively. Patients were divided into underweight (<18.5kg/m2), normal weight (20-24.9 kg/m2), overweight/obese (≥25kg/m2) based on BMI, who were followed for 180 days for all-cause death. Results 127 consecutive patients with obtained data were finally enrolled (60±15 years, 70% males), including 20 (15.75%) underweight, 75 (59.06%) normal weight, and 32 (25.19%) overweight/obese. As shown in Table 1, As shown in Table 1, the body fat mass and muscle mass varied with BMI, though they showed a different rate of change. Compared with patients with normal weight, the fat and muscle mass in those overweight/obese was 81.8% and 12.4% higher; whereas it was 39.2%, and 23.1% lower in those underweight, respectively. As results, the overweight/obesity group had a higher PBF, but a lower percentage of PBM. Consequently, the overweight/obesity group had a higher PBF, a lower PBM and a decreased muscle mass/fat mass ratio (MFR); conversely, the underweight group had a lower PBF, a higher PBM and an increased MFR. During 180-day follow-up, 17 (13.4%) died for all cause. By Cox regression analysis, MFR independently predict all-cause death (HR, 1.273, 95% CI: 1.056-1.534, p=0.011) after adjusted for age, BMI, ejection fraction and NT-proBNP. Of note, in patients with normal weight, Receiver Operating Characteristic (ROC) curve analysis suggested the capability of MFR (AUC=0.780, p=0.003) as well as its superiority over BMI in association with all-cause mortality (Figure 1). Conclusions The change of body fat or muscle mass with BMI displayed a different rate in HF patients, so that the MFR was higher in underweight but lower in overweight/obese when compared to normal weight group. Apart from BMI, MFR emerged as a predictor of mortality, in particular in patients with normal weight.