The development of immunodeficient mouse xenograft models has greatly facilitated the investigation of some human hematopoietic malignancies, but application of this approach to the myelodysplastic syndromes (MDSs) has proven difficult. We now show that cells from most MDS patients (including all subtypes) repopulate nonobese diabetic-severe combined immunodeficient (scid)/scid-beta2 microglobulin null (NOD/SCID-beta2m(-/-)) mice at least transiently and produce abnormal differentiation patterns in this model. Normal marrow transplants initially produce predominantly erythroid cells and later predominantly B-lymphoid cells in these mice, whereas most MDS samples produced predominantly granulopoietic cells. In 4 of 4 MDS cases, the regenerated cells showed the same clonal markers (trisomy 8, n = 3; and 5q-, n = 1) as the original sample and, in one instance, regenerated trisomy 8(+) B-lymphoid as well as myeloid cells were identified. Interestingly, the enhanced growth of normal marrow obtained in NOD/SCID-beta2m(-/-) mice engineered to produce human interleukin-3, granulocyte-macrophage colony-stimulating factor, and Steel factor was seen only with 1 of 7 MDS samples. These findings support the concept that human MDS originates in a transplantable multilineage hematopoietic stem cell whose genetic alteration may affect patterns of differentiation and responsiveness to hematopoietic growth factors. They also demonstrate the potential of this new murine xenotransplant model for future investigations of MDS.