Background: Axicabtagene ciloleucel is an autologous anti-CD19 chimeric antigen receptor (CAR) T-cell therapy. In the previous analysis of the ZUMA-1 registrational study, with a median follow-up of 15·4 months (IQR 13·7–17·3), 89 (82%) of 108 assessable patients with refractory large B-cell lymphoma treated with axicabtagene ciloleucel achieved an objective response, and complete responses were noted in 63 (58%) patients. Here we report long-term activity and safety outcomes of the ZUMA-1 study. Methods: ZUMA-1 is a single-arm, multicentre, registrational trial at 22 sites in the USA and Israel. Eligible patients were aged 18 years or older, and had histologically confirmed large B-cell lymphoma—including diffuse large B-cell lymphoma, primary mediastinal B-cell lymphoma, and transformed follicular lymphoma—according to the 2008 WHO Classification of Tumors of Hematopoietic and Lymphoid Tissue; refractory disease or relapsed after autologous stem-cell transplantation; an Eastern Cooperative Oncology Group performance status of 0 or 1; and had previously received an anti-CD20 monoclonal antibody containing-regimen and an anthracycline-containing chemotherapy. Participants received one dose of axicabtagene ciloleucel on day 0 at a target dose of 2 × 10 6 CAR T cells per kg of bodyweight after conditioning chemotherapy with intravenous fludarabine (30 mg/m 2 body-surface area) and cyclophosphamide (500 mg/m 2 body-surface area) on days −5, −4, and −3. The primary endpoints were safety for phase 1 and the proportion of patients achieving an objective response for phase 2, and key secondary endpoints were overall survival, progression-free survival, and duration of response. Pre-planned activity and safety analyses were done per protocol. ZUMA-1 is registered with ClinicalTrials.gov, number NCT02348216. Although the registrational cohorts are closed, the trial remains open, and recruitment to extension cohorts with alternative endpoints is underway. Findings: Between May 19, 2015, and Sept 15, 2016, 119 patients were enrolled and 108 received axicabtagene ciloleucel across phases 1 and 2. As of the cutoff date of Aug 11, 2018, 101 patients assessable for activity in phase 2 were followed up for a median of 27·1 months (IQR 25·7–28·8), 84 (83%) had an objective response, and 59 (58%) had a complete response. The median duration of response was 11·1 months (4·2–not estimable). The median overall survival was not reached (12·8–not estimable), and the median progression-free survival was 5·9 months (95% CI 3·3–15·0). 52 (48%) of 108 patients assessable for safety in phases 1 and 2 had grade 3 or worse serious adverse events. Grade 3 or worse cytokine release syndrome occurred in 12 (11%) patients, and grade 3 or worse neurological events in 35 (32%). Since the previous analysis at 1 year, additional serious adverse events were reported in four patients (grade 3 mental status changes, grade 4 myelodysplastic syndrome, grade 3 lung infection, and two episodes of grade 3 bacteraemia), none of which were judged to be treatment related. Two treatment-related deaths (due to haemophagocytic lymphohistiocytosis and cardiac arrest) were previously reported, but no new treatment-related deaths occurred during the additional follow-up. Interpretation: These 2-year follow-up data from ZUMA-1 suggest that axicabtagene ciloleucel can induce durable responses and a median overall survival of greater than 2 years, and has a manageable long-term safety profile in patients with relapsed or refractory large B-cell lymphoma. Funding: Kite and the Leukemia & Lymphoma Society Therapy Acceleration Program.