BACKGROUND: Lisocabtagene maraleucel (liso-cel) is an autologous, CD19-directed, chimeric antigen receptor (CAR) T-cell product. We aimed to assess the activity and safety of liso-cel in patients with relapsed or refractory large B-cell lymphomas. METHODS: We did a seamless design study at 14 cancer centres in the USA. We enrolled adult patients (aged ≥18 years) with relapsed or refractory large B-cell lymphomas. Eligible histological subgroups included diffuse large B-cell lymphoma, high-grade B-cell lymphoma with rearrangements of MYC and either BCL2, BCL6, or both (double-hit or triple-hit lymphoma), diffuse large B-cell lymphoma transformed from any indolent lymphoma, primary mediastinal B-cell lymphoma, and follicular lymphoma grade 3B. Patients were assigned to one of three target dose levels of liso-cel as they were sequentially tested in the trial (50 × 10 CAR T cells [one or two doses], 100 × 10 CAR T cells, and 150 × 10 CAR T cells), which were administered as a sequential infusion of two components (CD8 and CD4 CAR T cells) at equal target doses. Primary endpoints were adverse events, dose-limiting toxicities, and the objective response rate (assessed per Lugano criteria); endpoints were assessed by an independent review committee in the efficacy-evaluable set (comprising all patients who had confirmed PET-positive disease and received at least one dose of liso-cel). This trial is registered with ClinicalTrials.gov, NCT02631044. FINDINGS: Between Jan 11, 2016, and July 5, 2019, 344 patients underwent leukapheresis for manufacture of CAR T cells (liso-cel), of whom 269 patients received at least one dose of liso-cel. Patients had received a median of three (range 1-8) previous lines of systemic treatment, with 260 (97%) patients having had at least two lines. 112 (42%) patients were aged 65 years or older, 181 (67%) had chemotherapy-refractory disease, and seven (3%) had secondary CNS involvement. Median follow-up for overall survival for all 344 patients who had leukapheresis was 18·8 months (95% CI 15·0-19·3). Overall safety and activity of liso-cel did not differ by dose level. The recommended target dose was 100 × 10 CAR T cells (50 × 10 CD8 and 50 × 10 CD4 CAR T cells). Of 256 patients included in the efficacy-evaluable set, an objective response was achieved by 186 (73%, 95% CI 66·8-78·0) patients and a complete response by 136 (53%, 46·8-59·4). The most common grade 3 or worse adverse events were neutropenia in 161 (60%) patients, anaemia in 101 (37%), and thrombocytopenia in 72 (27%). Cytokine release syndrome and neurological events occurred in 113 (42%) and 80 (30%) patients, respectively; grade 3 or worse cytokine release syndrome and neurological events occurred in six (2%) and 27 (10%) patients, respectively. Nine (6%) patients had a dose-limiting toxicity, including one patient who died from diffuse alveolar damage following a dose of 50 × 10 CAR T cells. INTERPRETATION: Use of liso-cel resulted in a high objective response rate, with a low incidence of grade 3 or worse cytokine release syndrome and neurological events in patients with relapsed or refractory large B-cell lymphomas, including those with diverse histological subtypes and high-risk features. Liso-cel is under further evaluation at first relapse in large B-cell lymphomas and as a treatment for other relapsed or refractory B-cell malignancies. FUNDING: Juno Therapeutics, a Bristol-Myers Squibb Company.