Background: Treatment options for melanoma, which has the highest mutation burden among common cancers, has proliferated in the past decade. Genomic profiling has becoming essential to clinical decision making. However, limited studies have interrogated the genomic landscape of Chinese melanoma patients. We also investigated the correlation between tumor mutation burden (TMB) and clinical outcomes of immunotherapy (IO). Methods: In this study, we retrospectively surveyed the genomic profiling of primary tumors of 81 (40 males, 41 females) metastatic Chinese melanoma patients with a median age of 52, using a panel consisting of 295 cancer-related genes, spanning 2.02MB of human genome. Patients used IO as first line treatment (n = 25) were enrolled for survival analyses. Results: In this cohort, 15, 24 and 42 were acral, mucosal and cutaneous melanoma, respectively. Collectively, we identified 1,114 mutations, spanning 248 genes, with BRAF, MYC and NBN being the most frequently mutated genes, occurring in 40%, 27% and 21% of patients, respectively. Mutation spectrum varied significantly by subtypes. BRAF (57%) and LRP1B (26%) were the most frequently mutated genes in cutaneous melanoma (CM). KIT and NRAS, reported to be frequently mutated in CM, each occurred in only 12% patients in this cohort. MYC amplification was the most commonly seen mutation in acral and mucosal melanoma (MM). Other frequent mutations in MM included: NBN (38%) RUNX1T1(29%) and TP53 (29%). In acral melanoma (AM), CCND1, FGF3/19, NF1and NBN were frequently mutated. It is interesting to note that no TP53 mutation was observed in AM. AM and MM had significantly more CNVs than CM. Of the 25 patients underwent IO, our data revealed a positive correlation between TMB and PFS (p = 0.007). Such correlation also exited in each subtype. Furthermore, we derived a cutoff of 15, which can effectively distinguish clinical response. Patients with TMB > 15 mut/Mb had a significantly longer PFS than patients harboring TMB < 15 mut/Mb (P = 0.02). Patients with CM had a longer PFS than patients with AM or MM (p = 0.018). No correlation between PDL1 expression and PFS was observed. Conclusions: Our study revealed a distinctive mutation landscape for each subtype. Furthermore, we also revealed a positive correlation between TMB and PFS as well as a lack of correlation between PDL1 expression and PFS.