Cell therapy using T cells has revolutionized medical care in recent years but limitations are associated with the difficulty of genome editing of the cells, the production of a sufficient number of cells and standardization of the product. Human pluripotent stem cells (hPSCs) can self-renew and differentiate into T cells to provide a standardized homogenous product of defined origin in indefinite quantity, therefore they are of great potential to alleviate limitations of therapeutic T cell production. The differentiation of hPSCs takes place in two steps: first the induction of hematopoietic stem/progenitor cells (HSPCs), then the induction of lymphopoiesis by Notch signaling. However, the differentiation of T cells from hPSCs can be difficult and lack reproducibility. One parameter that needs to be better assessed is the potential of DLL1 vs. DLL4 ligands of the Notch pathway to induce T cells. In addition, culture of hPSCs is labor-intensive and not compatible with GMP production, especially when they are cultured on feeder cells. Thus, the definition of a robust GMP-compatible differentiation protocol from hPSCs cultured in feeder-free conditions would increase the accessibility to off-the-shelf hematopoietic and T cell progenitors derived from hPSCs. In this article, we describe an efficient, rapid and reproducible protocol for the generation of hematopoietic and T cell progenitors in two steps: (1) generation of HSPCs from embryoid bodies (EB) in serum free medium and GMP-compatible feeder-free systems, (2) directed differentiation of hPSC-derived HSPCs into T-cell progenitors in the presence of bone marrow stromal cells expressing Notch-ligands OP9-DLL1 vs. OP9-DLL4.