Natural regulatory T cells (Tregs) may have a great therapeutic potential to induce tolerance in allogeneic cells and organ transplantations. In mice, we showed that alloantigen-specific Tregs (spe-Tregs) were more efficient than polyclonal Tregs (poly-Tregs) in controlling graft-versus-host disease (GVHD). Here we describe a clinical-grade compliant method for generating human spe-Tregs. Tregs were enriched from leukapheresis products with anti-CD25 immunomagnetic beads, primed twice by allogeneic mature monocyte-derived dendritic cells (mDCs), and cultured during 3 weeks in medium containing interleukin 2 (IL-2), IL-15, and rapamycin. After 3 weeks of culture, final cell products were expanded 8.3-fold from the initial CD25(+) purifications. Immunophenotypic analyses of final cells indicate that they were composed of 88 ± 2.6% of CD4(+) T cells, all expressing Treg-specific markers (FOXP3, Helios, GARP, LAP, and CD152). Spe-Tregs were highly suppressive in vitro and also in vivo using a xeno-GVHD model established in immunodeficient mice. The specificity of their suppressive activity was demonstrated on their ability to significantly suppress the proliferation of autologous effector T cells stimulated by the same mDCs compared to third-party mDCs. Our data provide evidence that functional alloantigen Tregs can be generated under clinical-grade compliant conditions. Taking into account that 130 × 10(6) CD25(+) cells can be obtained at large scale from standard leukapheresis, our cell process may give rise to a theoretical final number of 1 × 10(9) spe-Tregs. Thus, using our strategy, we can propose to prepare spe-Tregs for clinical trials designed to control HLA-mismatched GVHD or organ transplantation rejection.