There is no consensus about the toxicity of erythropoiesis-stimulating agents among hemodialysis patients. We aimed to calculate the risk of death according to anemia control and erythropoietin (EPO) dosing among end-stage renal disease patients undergoing hemodialysis. We retrospectively studied 156 end-stage renal disease patients on hemodialysis from a single renal unit during 12 months. Participants were classified according to anemia control into four groups: excellent (A), good (B), moderate (C) and bad (D) control. They were also classified according to EPO dosing into two groups: usual and high EPO dosing. The Cox proportional hazards regression model, adjusted for the difference in age, sex, time on dialysis, comorbidity, albumin, and Kt/V index, was performed to calculate the risk of death according to anemia control and EPO dosing profiles. Multivariate analysis by backward stepwise logistic regression was used to calculate the risk of death according to the variables that differed in the comparison between survivors and nonsurvivors. The hazard ratio of death was not significant according to anemia control profile C/D vs. A/B, but hazard ratio was 2.967 (95% confidence interval [CI] = 1.132-7.777; P = 0.027) for high EPO dosing profile patients. The multivariate analysis showed comorbidity (odds ratio [OR] = 8.958; 95% CI = 2.843-26.223; P < 0.001], high EPO dosing profile (OR = 5.172; 95% CI = 1.663-16,081; P = 0.005), age (OR = 1.056; 95% CI = 1.020-1.094; P = 0.002), and mean hemoglobin (OR = 0.435; 95% CI = 0.267-0.709; P = 0.001) to be predictive of death. Even though we cannot conclude that mortality risk is due to EPO toxicity, hemodialysis patients using high EPO dosing must be seen as at risk.