<i>Background: </i>Recombinant human erythropoietin (rHuEPO) is recommended pre-dialysis to correct the anemia of chronic kidney disease. This study evaluated the impact of pre-dialysis rHuEPO on mortality in incident end-stage renal disease (ESRD) patients with varying levels of pre-ESRD care. <i>Methods:</i> The study included 15,807 individuals whose exposure to rHuEPO was determined from HCFA 2728 forms. <i>Results:</i> Median follow-up after starting dialysis was 32.8 months. Pre-ESRD rHuEPO use occurred in only 3,994 (25.3%) subjects and was more common in individuals with insurance, currently employed, started on outpatient dialysis, and initiated on peritoneal dialysis. During the study, 8,608 (54.5%) patients died. The risk of death was lower for rHuEPO-treated patients versus non-treated (relative risk 0.87, 95% CI 0.82–0.92). The survival benefit with rHuEPO was greatest early after dialysis initiation (relative risk at 1 vs. 7 years post-dialysis 0.73, 95% CI 0.66–0.80 vs. 0.87, 95% CI 0.82–0.92, respectively), did not vary across several surrogates for quality of care, and was greatest in those with the highest achieved hematocrit pre-ESRD. <i>Conclusion:</i> Pre-dialysis rHuEPO confers a survival benefit that depends on achieved hematocrit and diminishes post-dialysis, but is independent of several surrogates for quality of care except for insurance status pre-ESRD.