The diagnostic criteria of acute erythroid leukemias (AEL) has been revised by WHO in 2001. The National Cancer Institute (NCI) published a set of standardized diagnostic and response criteria for acute myeloid leukemia in 1990, which was revised in 2003. The aim of the present study was to establish the best criteria for therapeutic response assessment in the newly classified AEL and evaluate patient outcomes. Fifty-two patients with AEL as defined by the new WHO criteria were evaluated in this study. The following seven indices for therapeutic response assessment were evaluated: (i) NCI criteria (myeloblast percentage among total nucleated cells (TNC) and cellularity); (ii) myeloblast percentage among non-erythroid cells (NEC) and cellularity; (iii) erythroid series percentage among TNC; (iv) pronormoblast percentage among erythroid cells; (v) ratio of pronormoblasts and blasts; (vi) maturation arrest index; and (vii) disappearance of erythroid dysplasia. Complete remission (CR) patients with <5% of myeloblast/NEC (NEC-CR) showed significantly longer overall survival periods (mean 55.8 months) compared to CR patients with >5% myeloblast/NEC (mean 11.7 months, P = 0.006). NEC-CR patients also had longer event-free survival (median 16.4 months) compared to patients with >5% and <20% of myeloblast/NEC (median 6.2 months) (P = 0.044). The other indices for therapeutic response assessment are not significant for predictability of relapse and outcomes. Therefore, we recommend that the myeloblast percentage among NEC be used instead of myeloblast percentage among TNC for therapeutic response assessment in AEL.