The adoptive transfer of autologous, long-lived, gene-repaired T cells is a promising way to treat inherited T-cell immunodeficiencies. However, adoptive T-cell therapies require a large number of T cells to be manipulated and infused back into the patient. This poses a challenge in primary immunodeficiencies that manifest early in childhood and where only small volumes of blood samples may be available. Our protocol describes the ex vivo expansion of potentially long-lived human T memory stem cells (T_SCM), starting from a limited number of peripheral blood mononuclear cells (PBMCs). Using the perforin gene as an example, we provide detailed instructions for precise gene repair of human T cells and the expansion of T_SCM. The efficiency of precise gene repair can be increased by suppressing unintended non-homologous end-joining (NHEJ) events. Our protocol yields edited T-cell populations that are ready for phenotyping, genome-wide off-target analysis, and functional characterization.