Adoptive cell therapy with T cells reprogrammed to express chimeric antigen receptors (CAR-T cells) has been highly successful in patients with hematological neoplasms. However, its therapeutic benefits have been limited in solid tumor cases. Even those patients who respond to this immunotherapy remain at risk of relapse due to the short-term persistence or non-expansion of CAR-T cells; moreover, the hostile tumor microenvironment (TME) leads to the dysfunction of these cells after reinfusion. Some research has shown that, in adoptive T-cell therapies, the presence of less differentiated T-cell subsets within the infusion product is associated with better clinical outcomes. Naive and memory T cells persist longer and exhibit greater antitumor activity than effector T cells. Therefore, new methods are being studied to overcome the limitations of this therapy to generate CAR-T cells with these ideal phenotypes. In this paper, we review the characteristics of T-cell subsets and their implications in the clinical outcomes of adoptive therapy with CAR-T cells. In addition, we describe some strategies developed to overcome the reduced persistence of CAR T-cells and alternatives to improve this therapy by increasing the expansion ability and longevity of modified T cells. These methods include cell culture optimization, incorporating homeostatic cytokines during the expansion phase of manufacturing, modulation of CAR-T cell metabolism, manipulating signaling pathways involved in T-cell differentiation, and strategies related to CAR construct designs.