Since the discovery of the immunomodulation property of mesenchymal stem cells (MSCs) about a decade ago, it has been extensively investigated whether MSCs can be used for the treatment of immune-related diseases, such as graft-versus-host disease (GvHD). However, how to evaluate the efficacy of human MSCs for the clinical trial is still unclear. We used an MHC-mismatched model of GvHD (B6 into BALB/c). Surprisingly, the administration of the human MSCs (hMSCs) could reduce the GvHD-related mortality of the mouse recipients and xenogeneically inhibit mouseT-cell proliferation and IFN-gamma productionin vitro. We recently established a new protocol for the isolation of a homogeneous population of MSCs called subfractionation culturing methods (SCM), and established a library of clonal MSC lines. Therefore, we also investigated whether MSCs isolated by the conventional gradient centrifugationmethod (GCM) and SCM show different efficacy in vivo. Intriguingly, clonal hMSCs (hcMSCs) isolated by SCM showed better efficacy than hMSCs isolated by GCM. Based on these results, the MHC-mismatched model of GvHD may be useful for evaluating the efficacy of human MSCs before the clinical trial. The results of this study suggest that different MSC lines may show different efficacy in vivo and in vitro.