Muscle-based gene therapy and tissue engineering hold great promise for improving bone healing. However, the relative advantage of muscle-derived stem cells (MDSCs) or primary muscle-derived cells (MDCs) remains to be defined. We compared the ability of MDSCs and different subpopulations of MDCs (PP1 and PP3) to induce bone formation via ex vivo gene therapy. We were able to efficiently transduce the MDSCs and all the other evaluated populations of MDCs (efficiency of transduction = approximately 80%) by using a retroviral vector expressing human bone morphogenetic protein 4 (BMP4). All the transduced cell populations secreted high levels of BMP4 (140-300 ng/10(6) cells/24 h), but the MDSCs differentiated toward the osteogenic lineage more effectively than did the other muscle cell populations, as indicated by the expression of alkaline phosphatase, an early osteogenic marker. von Kossa staining indicated that mineralized bone formed as early as 7 days after implantation of any of the BMP4-expressing cell populations into immunocompetent syngeneic mice; however, MDSCs expressing BMP4 produced significantly more bone than did the other MDC populations, as evidenced by both histomorphometry and biochemical analysis. Further investigation revealed that MDSCs expressing BMP4 persisted for a significantly longer period of time at the bone forming sites than did the other BMP4-expressing MDC populations. Additionally, MDSCs expressing BMP4 triggered a smaller infiltration of CD4 lymphocytes within the bone forming areas than did the other MDC populations expressing BMP4. Finally, we demonstrated that MDSCs expressing BMP4 can heal a critical-sized skull bone defect in immunocompetent mice. In summary, this study shows that MDSCs are better than primary MDCs for use as cellular vehicles in BMP4-based ex vivo gene therapy to improve bone healing. The advantage of MDSCs may be attributable, at least in part, to their lower immunogenicity and higher capacity for in vivo survival.