A 26-year-old ABO-O positive patient with aplastic anemia received a bone marrow transplant from his genotypically HLA identical, but ABO-A positive, brother. Engraftment of myeloid and megakaryocytic lineages occurred within 4 weeks but pure red cell aplasia and transfusion dependent anemia persisted for 160 days. The authors postulated that the failure of erythropoiesis was due to a high titer of anti- A isohemagglutinins. They tested this hypothesis with clonal cell cultures and flow cytometric analysis of ABO antigen expression by colony forming cells in vitro. During the period of prolonged red cell aplasia, the patient had normal numbers (85 ± 1 2 per 106cells) of circulating donor derived, burst forming units-erythroid (BFU-E). Immunophenotypic analysis of erythroid burst colonies derived from culture of the patient's bone marrow cells showed that 91 ± 5% of 274 nucleated red cells were A-antigen positive, confirming full donor engraftment. Autologous plasma and complement added on day 1 of culture did not affect the colony growth (82.5 ±15 per 106cells). However, when the addition of complement was delayed until day 7 of culture, there was 90% inhibition of BFU-E (7.5 ± 5 per 106cells) compared to controls (p<0.0004). Based on this, the authors propose a model for expression of ABO antigens during erythropoiesis, in which BFU-E do not express ABO antigens but their progeny do. The data support the hypothesis that the mechanism of prolonged pure red cell aplasia after ABO-incompatible bone marrow transplantation is complement mediated immune destruction of erythroid progenitors past the stage of BFU-E in differentiation.