Abstract Background Patients with heart failure (HF) often have elevated endogenous erythropoietin (EPO) levels, which are associated with disease severity and prognosis. Previous studies have indicated that in the special population elevated endogenous EPO levels are associated with HF and cardiovascular death. However, whether endogenous EPO levels are associated with the risk of HF in the general population remains to be elucidated. Purpose Our study aims to investigate the association between endogenous EPO levels and the risk of HF admission in the general population. Methods We conducted a prospective cohort study on subjects from a Beijing community in China who had no history of stroke or myocardial infarction. Endogenous EPO levels were measured at baseline for all participants. The endpoint of interest was admission with HF. Cox proportional hazard regression models were used to analyze the relationship between endogenous EPO levels and the occurrence of the endpoint. Results A total of 4870 participants were included. Among all participants, the mean (±SD) age was 56.72 ± 8.78 years, with 1802 (37.0%) being men. The median (interquartile range, IQR) EPO level was 13.19 (9.47-18.20) IU/L. During a mean 9.57-year follow-up, a total of 176 (3.6%) subjects experienced heart failure. Subjects were grouped in quintiles according to baseline endogenous EPO levels. Higher and lower endogenous EPO levels were both associated with increased risk of HF (Figure 1A). The Kaplan–Meier curves showed significant differences in cumulative hazards of heart failure among EPO quintiles (log-rank test: P<0.05) (Figure 1B). After adjusting for covariates, including demographics, traditional risk factors for cardiovascular disease, renal function, and hemoglobin levels, and using the middle quintile (Q3) (11.72-14.94IU/L) of EPO levels as a reference, we found that the highest quintile (Q5) (>19.86 IU/L) was associated with a 159% increased risk of HF admission (HR 2.59, 95% CI 1.57-4.28, P = 0.0002). Similarly, the lowest quintile (Q1) (<8.66 IU/L) was associated with a 94% increased risk (HR 1.94, 95% CI 1.11-3.41, P = 0.0210) (Table 1). Conclusion There exists a "U-shaped" relationship between endogenous EPO levels and HF admission in the general population. Both higher and lower endogenous EPO levels were associated with an increased risk of HF admission in the community population. These findings differ from previous studies, so further research may be needed to explore the underlying pathophysiological mechanisms of endogenous EPO and HF.