BACKGROUND: Immune checkpoint inhibitors (ICIs) represent an appealing treatment for patients with advanced urothelial cancer (aUC) and a poor performance status (PS). However, the benefit of ICIs for patients with a poor PS remains unknown. It was hypothesized that a poor Eastern Cooperative Oncology Group (ECOG) PS (≥2 vs 0-1) would correlate with shorter overall survival (OS) in patients receiving ICIs. METHODS: In this retrospective cohort study, clinicopathologic, treatment, and outcome data were collected for patients with aUC who were treated with ICIs at 18 institutions (2013-2019). The overall response rate (ORR) and OS were compared for patients with an ECOG PS of 0 to 1 and patients with an ECOG PS ≥ 2 at ICI initiation. The association between a new ICI in the last 30 and 90 days of life (DOL) and death location was also tested. RESULTS: Of the 519 patients treated with ICIs, 395 and 384 were included in OS and ORR analyses, respectively, with 26% and 24% having a PS ≥ 2. OS was higher in those with a PS of 0 to 1 than those with a PS ≥ 2 who were treated in the first line (median, 15.2 vs 7.2 months; hazard ratio [HR], 0.62; P = .01) but not in subsequent lines (median, 9.8 vs 8.2 months; HR, 0.78; P = .27). ORRs were similar for patients with a PS of 0 to 1 and patients with a PS ≥ 2 in both lines. Of the 288 patients who died, 10% and 32% started ICIs in the last 30 and 90 DOL, respectively. ICI initiation in the last 30 DOL was associated with increased odds of death in a hospital (odds ratio, 2.89; P = .04). CONCLUSIONS: Despite comparable ORRs, ICIs may not overcome the negative prognostic role of a poor PS, particularly in the first-line setting, and the initiation of ICIs in the last 30 DOL was associated with hospital death location.