The peripheral Foxp3 + Treg pool consists of naturally arising Treg (nTreg) and adaptive Treg cells (iTreg). It is well known that naive CD4 + T cells can be readily converted to Foxp3 + iTreg in vitro, and memory CD4 + T cells are resistant to conversion. In this study, we investigated the induction of Foxp3 + T cells from various CD4 + T-cell subsets in human peripheral blood. Though naive CD4 + T cells were readily converted to Foxp3 + T cells with TGF-β and IL-2 treatment in vitro, such Foxp3 + T cells did not express the memory marker CD45RO as do Foxp3 + T cells induced in the peripheral blood of Hepatitis B Virus (HBV) patients. Interestingly, a subset of human memory CD4 + T cells, defined as CD62L + central memory T cells, could be induced by TGF-β to differentiate into Foxp3 + T cells. It is well known that Foxp3 + T cells derived from human CD4 +CD25 - T cells in vitro are lack suppressive functions. Our data about the suppressive functions of CD4 +CD62L + central memory T cell-derived Foxp3 + T cells support this conception, and an epigenetic analysis of these cells showed a similar methylation pattern in the FOXP3 Treg-specific demethylated region as the naive CD4 + T cell-derived Foxp3 + T cells. But further research showed that mouse CD4 + central memory T cells also could be induced to differentiate into Foxp3 + T cells, such Foxp3 + T cells could suppress the proliferation of effector T cells. Thus, our study identified CD4 +CD62L + central memory T cells as a novel potential source of iTreg.