The use of immunotherapy for treating advanced cancer has developed dramatically in recent years. Transferring T cells encoding tumour-specific chimeric antigen receptors (CAR) has shown great promise in treating haematological cancers but has been less successful in patients with solid tumours. Although partial remissions were observed in several clinical trials for solid cancer, most patients were stable or in disease progression at follow up. Furthermore, CAR T cells have been reported to expand less well and fail to persist after infusion to these patients. Therefore, the optimal production of long-lived and highly functional CAR T cells is clearly required for the successful application of this immunotherapy to solid tumours. Here, we investigate the ability of T lymphocytes derived from solid cancer patients receiving chemotherapy to result in the generation of satisfactory CAR T-cell products. we hypothesize that on-treatment blood specimens of solid cancer patients may contain a high proportion of central memory T cells and thus represent the optimal starting population for manufacturing highly functional and effective CAR T cells. To test this hypothesis, CAR T cells are produced using leukocyte populations isolated from glioblastoma, advanced pancreatic cancer and metastatic colorectal cancer patients at three-time points: before chemotherapy, during chemotherapy and at relapse. Two types of starting population,unsorted PBMCs and magnetically sorted CD4+ and CD8+ T cells, are compared for their transduction efficiency, expansion, and immune phenotype. Initial results have shown that on- treatment CAR T cells proliferate strongly and sorted T cells have a higher transduction efficiency. This study will determine of T cells harvested from cancer patientsat different stages of their disease can generate therapeutically useful yields of CAR T cells.