Background: Axi-cel, an anti-CD19 CART cell therapy, achieved 83% ORR, 58% CR rate, with 39% PFS at 2 years in patients (pts) with relapsed refractory large B-cell lymphoma (LBCL) on the ZUMA-1 study (Locke, Lancet Oncology 2018). Data from a 17-center consortium showed response rates were similar in 274 pts treated with commercial axi-cel (Nastoupil, ASH 2018). Here, we performed retrospective analysis of outcomes in pts with progressive disease (PD) after axi-cel. Methods: Response status was determined by Cheson 2014 and reported as date of radiologic relapse. 274 pts were infused by December 26, 2018 with maximum follow-up of 14 months; 116 pts had PD as of Feb 1, 2019. Twelve sites provided additional data, detailing 85% of PD pts (n=99) with a median time to relapse of 54 days (IQR 16-120). Results: Pre axi-cel pt characteristics: median lines of therapy were 4 (range 2-11), 86% were Stage III/IV and 22% were ECOG >1. Following relapse, 60% (n=61) were biopsied and 70% (43/61) had CD19 expression measured. Thirty percent (13/43) were CD19 negative by: IHC (3/13), flow (2/13) or both (8/13). Seventy percent (n=71) received salvage therapy for PD. Median lines of salvage therapy was 1 (range 0-4). The most common therapies were Lenalidomide-based (30%), checkpoint inhibitors (30%), chemotherapy (20%) and radiation (10%). First salvage therapy ORR by regimen: checkpoint inhibitors = 24%, lenalidomide regimens = 20% and chemotherapy = 11%. One patient received allogeneic transplant. Twelve pts enrolled on clinical trial, with one receiving 2^nd CAR-T. Median OS following relapse was 108 days (95% LCL 71). Nineteen pts relapsed <3 months after axi-cel and did not receive therapy with median OS 17 days (95% CI 7-49); 33 pts relapsed <3 months and received therapy with 114 day median OS (95% LCL 82). In contrast, 30 pts who relapsed >3 months post axi-cel and received therapy had estimated median OS >220 days (95% LCL 81). Conclusions: Patients with LBCL relapsing less than 3 months following axi-cel have extremely poor outcomes supporting the development of novel therapies. Therapy for relapse >3 months appears promising. JYS and SD contributed equally; APR, DBM and BTH contributed equally.