The transplantation of polymer encapsulated myoblasts genetically engineered to secrete erythropoietin (Epo) may obviate the need for repeated parenteral administration of recombinant Epo as a treatment for chronic renal failure, cancer or AIDS-associated anemia. To explore this possibility, the human and mouse Epo cDNAs under the control of the housekeeping mouse PGK-1 promoter were transfected into mouse C2C12 myoblasts, which can be terminally differentiated upon exposure to low serum-containing media. Pools releasing 150 IU human Epo per 10(6) cells per day and 390 IU mouse Epo per 10(6) cells per day were selected. Polyether-sulfone (PES) capsules loaded with approximately 200,000 transfected myoblasts from these pools were implanted on the dorsal flank of DBA/2J, C3H and C57BL/6 mice. With human Epo secreting capsules, only a transient increase in the hematocrit occurred in DBA/2J mice, whereas no significant response was detected in C3H or C57BL/6 mice. On the contrary, all mice implanted with capsules releasing mouse Epo increased their hematocrit over 85% as early as 7 days after implantation and sustained these levels for at least 80 days. All retrieved implants released Epo and contained well preserved myoblasts. Moreover most capsules were surrounded by a neovascularization. Mice transplanted with nonencapsulated C2C12 cells releasing mouse Epo showed only a transitory elevation of their hematocrit reflecting the poor engraftment of injected myoblasts. These results indicate that polymer encapsulation of genetically engineered myoblasts is a promising approach for the long-term delivery of bioactive molecules, allowing the resolution of the shortcomings of free myoblast transfer.