Following recognition of the cognate antigen, naïve T cells differentiate in a diverse progeny of memory T cells which differ at the phenotypic, gene expression and metabolic level. These molecular differences are at the basis of discrete functionality, migratory capacity and persistence in the long-term. Such a division of labor benefits adoptive T cell transfer immunotherapy approaches for cancer and viral infections, as increased persistence and effector functions in vivo result in better control of the disease. Preclinical data suggest that early-differentiated T memory stem cells are the most powerful anti-tumor T cell population following adoptive transfer, but their paucity ex vivo limits translation to the clinic. Here, we describe a simple protocol to derive large numbers of T memory stem cell and effector CD8+ T cell subsets from highly-purified CD8+ naïve T cell precursors. The obtained cells can be studied in vitro to understand the molecular basis of human memory T cell differentiation, or, when redirected with T cell receptor or chimeric antigen receptor, potentially used in vivo to favour T cell reconstitution or to treat established tumors or chronic infections.