Uveal melanoma (UM), though a rare form of melanoma, is the most common intraocular tumor in adults. Conventional therapies of primary tumors lead to an excellent local control, but 50% of patients develop metastases, in most cases with lethal outcome. Somatic driver mutations that act on the MAP-kinase pathway have been identified, yet targeted therapies show little efficacy in the clinics. No drugs are currently available for the G protein alpha subunitsGNAQ and GNA11, which are the most frequent driver mutations in UM. Drugs targeting the YAP–TAZ pathway that is also activated in UM, the tumor-suppressor gene BRCA1 Associated Protein 1 (BAP1) and the Splicing Factor 3b Subunit 1 gene (SF3B1) whose mutations are associated with metastatic risk, have not been developed yet. Immunotherapy is highly effective in cutaneous melanoma but yields only poor results in the treatment of UM: anti-PD-1 and anti-CTLA-4 blocking antibodies did not meet the expectations except for isolated cases. Here, we discuss how the improved knowledge of the tumor microenvironment and of the cross-talk between tumor and immune cells could help to reshape anti-tumor immune responses to overcome the intrinsic resistance to immune checkpoint blockers of UM. We critically review the dogma of low mutational load, the induction of immune-suppressive cells, and the expression of alternative immune checkpoint molecules. We argue that immunotherapy might still be an option for the treatment of UM.