Chimeric antigen receptor (CAR)-modified T cells targeting CD19 demonstrate unparalleled responses in relapsed/refractory acute lymphoblastic leukemia (ALL)1-5, but toxicity, including cytokine-release syndrome (CRS) and neurotoxicity, limits broader application. Moreover, 40-60% of patients relapse owing to poor CAR T cell persistence or emergence of CD19- clones. Some factors, including the choice of single-chain spacer6 and extracellular7 and costimulatory domains8, have a profound effect on CAR T cell function and persistence. However, little is known about the impact of CAR binding affinity. There is evidence of a ceiling above which increased immunoreceptor affinity may adversely affect T cell responses9-11. We generated a novel CD19 CAR (CAT) with a lower affinity than FMC63, the high-affinity binder used in many clinical studies1-4. CAT CAR T cells showed increased proliferation and cytotoxicity in vitro and had enhanced proliferative and in vivo antitumor activity compared with FMC63 CAR T cells. In a clinical study (CARPALL, NCT02443831 ), 12/14 patients with relapsed/refractory pediatric B cell acute lymphoblastic leukemia treated with CAT CAR T cells achieved molecular remission. Persistence was demonstrated in 11 of 14 patients at last follow-up, with enhanced CAR T cell expansion compared with published data. Toxicity was low, with no severe CRS. One-year overall and event-free survival were 63% and 46%, respectively.