Abstract Background Patients with metastatic uveal melanoma (MUM) in the liver usually die within 1 year. The development of new treatments for MUM has been limited by the lack of diverse MUM cell lines and appropriate animal models. We previously reported that orthotopic xenograft mouse models established by direct injection of MUM cells into the liver were useful for the analysis associated with tumor microenvironment in the liver. However, considering that patients with UM metastasize to the liver hematogenously, direct liver injection model might not be suitable for investigation on various mechanisms of liver metastasis. Here, we aim to establish new orthotopic xenograft models via hematogenous dissemination of tumor cells to the liver, and to compare their characteristics with the hepatic injection model. We also determine if hepatic tumors could be effectively monitored with non-invasive live imaging. Methods tdtTomate-labeled, patient-derived MUM cells were injected into the liver, spleen or tail vein of immunodeficient NSG mice. Tumor growth was serially assessed with In Vivo Imaging System (IVIS) images once every week. Established hepatic tumors were evaluated with CT scan and then analyzed histologically. Results We found that splenic injection could consistently establish hepatic tumors. Non-invasive imaging showed that the splenic injection model had more consistent and stronger fluorescent intensity compared to the hepatic injection model. There were no significant differences in tumor growth between splenic injection with splenectomy and without splenectomy. The splenic injection established hepatic tumors diffusely throughout the liver, while the hepatic injection of tumor cells established a single localized tumor. Long-term monitoring of tumor development showed that tumor growth, tumor distribution in the liver, and overall survival depended on the number of tumor cells injected to the spleen. Conclusion We established a new orthotopic hepatic metastatic xenograft mouse model by splenic injection of MUM cells. The growth of orthotopic hepatic tumors could be monitored with non-invasive IVIS imaging. Moreover, we evaluated the therapeutic effect of a MEK inhibitor by using this model. Our findings suggest that our new orthotopic liver metastatic mouse model may be useful for preclinical drug screening experiments and for the analysis of liver metastasis mechanisms.