The effect of erythropoiesis-stimulating agent (ESA) on dialysis initiation in advanced chronic kidney disease (CKD) patients is not clear. We retrospectively analyzed the outcome of dialysis initiation in a stage 5 CKD cohort with ESA reimbursement limited to the maximal standardized monthly ESA dose equivalent to epoetin beta 20,000 U by the National Health Insurance program. Totally 423 patients were followed up for a median of 1.37 year. A time-dependent Cox regression model, adjusted for monthly levels of estimated glomerular filtration rate (eGFR) and hemoglobin, was constructed to investigate the association between ESA and outcome. The standardized monthly ESA dose in ESA users was 16,000 ± 3,900 U of epoetin beta. Annual changes of hemoglobin were -0.29 ± 2.19 and -0.99 ± 2.46 g/dL in ESA users and ESA non-users, respectively (P = 0.038). However, annual eGFR decline rates were not different between ESA users and non-users. After adjustment, ESA use was associated with deferred dialysis initiation (hazard ratio 0.63, 95% confidence interval 0.42-0.93, P = 0.021). The protective effect remained when the monthly ESA doses were incorporated. Our data showed that restricted use of ESA was safe and associated with deferred dialysis initiation in stage 5 CKD patients.