BACKGROUNDMucosal melanoma (MM) is a rare melanoma subtype with distinct biology and poor prognosis. Data on the efficacy of immune checkpoint inhibitors (ICIs) are limited. We determined the efficacy of ICIs in MM, analyzed by primary site and ethnicity/race. PATIENTS AND METHODSA retrospective cohort study from 25 cancer centers in Australia, Europe, USA and Asia was carried out. Patients with histologically confirmed MM were treated with anti-programmed cell death protein 1 (PD-1) ± ipilimumab. Primary endpoints were response rate (RR), progression-free survival (PFS), overall survival (OS) by primary site (naso-oral, urogenital, anorectal, other), ethnicity/race (Caucasian, Asian, Other) and treatment. Univariate and multivariate Cox proportional hazards model analyses were conducted. RESULTSIn total, 545 patients were included: 331 (63%) Caucasian, 176 (33%) Asian and 20 (4%) Other. Primary sites included 113 (21%) anorectal, 178 (32%) urogenital, 206 (38%) naso-oral and 45 (8%) other. Three hundred and forty-eight (64%) patients received anti-PD-1 and 197 (36%) anti-PD-1/ipilimumab. RR, PFS and OS did not differ by primary site, ethnicity/race or treatment. RR for naso-oral was numerically higher for anti-PD-1/ipilimumab 40%, 95% confidence interval (CI) 29% to 54% compared with anti-PD-1 (29%, 95% CI 21% to 37%). Thirty-five percent of patients who initially responded progressed. The median duration of response (mDoR) was 26 months (95% CI 18 months-not reached). Factors associated with short PFS were Eastern Cooperative Oncology Group (ECOG) performance status (PS) ≥3 (P < 0.01), lactate dehydrogenase (LDH) more than the upper limit of normal (ULN) (P = 0.01), lung metastases (P < 0.01) and ≥1 previous treatments (P < 0.01). Factors associated with short OS were ECOG PS ≥1 (P < 0.01), LDH >ULN (P = 0.03), lung metastases (P < 0.01) and ≥1 previous treatments (P < 0.01). CONCLUSIONSMM has poor prognosis. Treatment efficacy of anti-PD-1 ± ipilimumab was similar and did not differ by ethnicity/race. Naso-oral primaries had numerically higher response to anti-PD-1/ipilimumab, without difference in survival. The addition of ipilimumab did not show greater benefit over anti-PD-1 for other primary sites. In responders, mDoR was short and acquired resistance was common. Other factors, including site and number of metastases, were associated with survival.