Background & Aim Chimeric Antigen Receptor (CAR) T cell therapy has emerged as a treatment for systemic cancers such as acute lymphoblastic leukaemia and non-Hodgkin lymphoma with currently two products (Kymriah™ and Yescarta™) approved. In current studies, toxicity and long-term efficacy of CAR-T cells remain challenging. Contributing factors include phenotypic heterogeneity of T cells with low numbers of potent and long-lasting subsets such, as T Central Memory (TCM), and a high number of undesired terminally differentiated T cells. Most ex vivo expansion protocols routinely feed T cells with media containing a high concentration of glucose. The main hypothesis of the presented study is that culturing T cells in a glucose-deprived environment would limit differentiation and improve the efficacy through producing a TCM -enriched CAR-T product. Methods, Results & Conclusion The effects of altering the glucose concentration (0–25 mM) and the length of T cell exposure to glucose deprivation (0-8 days) on the growth and phenotypic composition of T cells were analysed. Our data showed that culturing T cells in a high concentration of glucose for the first 3 days and then switching to glucose-deprivation enhances TCM generation and suppresses excessive differentiation of T cells into terminal effector phenotype. The cytotoxic CD19-directed CAR-modified T cells generated under this condition retained their ability to degranulate and to produce inflammatory cytokines in vitro. Collectively, these results revealed the feasibility of this feeding regime to produce a CAR-T product enriched with more than 90% TCM. Thus potentially mitigating the toxicity associated with the presence of a large number of differentiated cells. In addition, the number of memory and effector cells can be simply altered by the amount of glucose added in the media during CAR-T manufacturing. This suggests a potential strategy to redirect T cell differentiation and control T cell composition during manufacturing.