Background— Studies suggest that in patients with heart failure (HF), high serum erythropoietin is associated with risk of recurrent HF and mortality. Trials of erythropoietin-stimulating agents in persons with kidney disease have also suggested an increased incidence of adverse clinical events. No large studies of which we are aware have evaluated the association of endogenous erythropoietin levels with clinical outcomes in the community-living older adults. Methods and Results— Erythropoietin concentration was measured in 2488 participants aged 70–79 years in the Health, Aging and Body Composition Study. Associations of erythropoietin with incident HF, coronary heart disease, stroke, mortality, and ≥30% decline in estimated glomerular filtration rate were examined using Cox proportional hazards and logistic regression over 10.7 years of follow-up. Mean (SD) age was 75 (3) years and median (quartile 1, quartile 3) erythropoietin was 12.3 (9.0, 17.2) mIU/mL. There were 503 incident HF events, and each doubling of serum erythropoietin was associated with a 25% increased risk of incident HF 1.25 (95% confidence interval 1.13, 1.48) after adjusting for demographics, prevalent cardiovascular disease, cardiovascular disease risk factors, kidney function, and serum hemoglobin. There was no interaction of serum erythropoietin with chronic kidney disease or anemia ( P >0.50). There were 330 incident coronary heart disease events, 161 strokes, 1112 deaths, and 698 outcomes of ≥30% decline in estimated glomerular filtration rate. Serum erythropoietin was not significantly associated with these outcomes. Conclusions— Higher levels of endogenous erythropoietin are associated with incident HF in older adults. Studies need to elucidate the mechanisms through which endogenous erythropoietin levels associate with specific outcomes.