Erythropoiesis progresses from stem cell expansion on stromal cells through the formation of an erythroblastic island. Our aim was to assess the feasibility of using human stromal cells for erythroid production and differentiation. When cord blood CD34+ cells were cocultured with telomerized human stromal cells (hTERT-stromal cells) for 2 weeks, the CD34+ cells and burst-forming units-erythroid (BFU-E) significantly expanded, and a few hematopoietic cells transmigrated below the stromal layer. When nonadherent hematopoietic progenitor cells that had expanded above the hTERT-stromal cells (group B) were collected and subjected to our erythroid-differentiation protocol, they differentiated into erythroblasts with a slight hemoglobin synthesis. When the few hematopoietic cells that had transmigrated below the stromal layer were expanded for an additional 2 to 6 weeks, they exhibited a cobblestone-like appearance, and a large amount of BFU-E clambered weekly from the underside of the stromal layer to above the stromal layer (group C). When the hematopoietic progenitor cells in group C were subjected to the erythroid-differentiation protocol, large numbers of mature erythroblasts (more than 300,000 times the initial CD34+ cell number) were produced. Our hTERT-stromal expansion protocol may contribute to the construction of a system for large-scale, long-term production of erythroid cells.