The molecular, histopathological, genomic and transcriptomic characteristics of uveal melanoma (UM) have identified four molecular subgroups with different clinical outcomes. Despite the improvements in UM classification and biological pathology, current treatments do not reduce the occurrence of metastasis. The development of effective adjuvant and metastatic therapies for UM has been slow and extremely limited. Preclinical models that closely resemble the molecular and genetic UM subgroups are essential for translating molecular findings into improved clinical treatment. In this review, we provide a retrospective view of the existing preclinical models used to study UM, and give an overview of their strengths and limitations. We review targeted therapy clinical trial data to evaluate the gap in the translation of preclinical findings to human studies. Reflecting on the current high attrition rates of clinical trials for UM, preclinical models that effectively recapitulate the human in vivo situation and/or accurately reflect the subtype classifications would enhance the translational impact of experimental data and have crucial implications for the advancement of personalised medicine.