In the FAB (French-American-British) and WHO (World Heath Organization) classifications, the blasts in erythroleukemia (M6a) are enumerated from the marrow nonerythroid rather than the total-nucleated cells. However, the method for blast calculation in erythroid-predominant myelodysplastic syndrome (erythroblasts>or=50%) is not specified either in the FAB or WHO classifications. We retrieved the files of 74 erythroid-predominant myelodysplastic syndrome patients (17% of all myelodysplastic syndrome) and 192 myelodysplastic syndrome controls (erythroblasts<50%). In erythroid-predominant myelodysplastic syndrome, by enumerating blasts from marrow nonerythroid cells rather than from total nucleated cells, 41 of 74 (55%) cases would be upgraded, either by disease subcategory or International Prognostic Scoring System. Importantly, the patients with <5% blasts demonstrated a superior survival to patients with >or=5% blasts (P=0.002); this distinction was lost when blasts were calculated from total-nucleated cells. Of cases with >or=5% blasts, cytogenetics rather than blast count correlated with survival. We conclude that in erythroid-predominant myelodysplastic syndrome, blast calculation as a proportion of marrow nonerythroid rather than total nucleated cells can better stratify patients into prognostically relevant groups.