Open questions in the pathogenesis of childhood acute lymphoblastic leukemia (ALL) are which hematopoietic cell is target of the malignant transformation and whether primitive stem cells contribute to the leukemic clone. Although good-prognosis ALL is thought to originate in a lymphoid progenitor, it is unclear if this applies to high-risk ALL. Therefore, immature CD34(+)CD19(-) bone marrow cells from 8 children with ALL/t(9;22) and 12 with ALL/t(4;11) were purified and analyzed by fluorescence in situ hybridization, reverse transcription-PCR (RT-PCR), and colony assays. Fifty-six percent (n = 8, SD 31%) and 68% (n = 12, SD 26%) of CD34(+)CD19(-) cells in ALL/t(9;22) and ALL/t(4;11), respectively, carried the translocation. In addition, 5 of 168 (3%) and 22 of 228 (10%) myeloerythroid colonies expressed BCR/ABL and MLL/AF4. RT-PCR results were confirmed by sequence analysis. Interestingly, in some patients with ALL/t(4;11), alternative splicing was seen in myeloid progenitors compared with the bulk leukemic population, suggesting that these myeloid colonies might be part of the leukemic cell clone. Fluorescence in situ hybridization analysis, however, shows that none of these myeloid colonies (0 of 41 RT-PCR-positive colonies) originated from a progenitor cell that carries the leukemia-specific translocation. Thus, leukemic, translocation-positive CD34(+)CD19(-) progenitor/stem cells that were copurified by cell sorting were able to survive in these colony assays for up to 28 days allowing amplification of the respective fusion transcripts by sensitive RT-PCR. In conclusion, we show that childhood high-risk ALL/t(9;22) and t(4;11) originate in a primitive CD34(+)CD19(-) progenitor/stem cell without a myeloerythroid developmental potential.