Melanoma is highly resistant to conventional chemotherapeutic agents and novel therapeutic approaches are needed. Current animal models of melanoma in animals are sub-optimal. The most commonly used homograft model is the B16 mouse melanoma. Evaluation of potential melanoma therapies with this model is limited by the inaccuracy of caliper measurement of subcutaneous tumors, of counting lung nodules in metastasis models, and the indirect nature of "survival" curves when studying brain metastases. We have developed and characterized an accurate, sensitive, and reproducible bioluminescent B16 melanoma model that allows for serial, real-time analyses of tumor burden in live mice. We demonstrate that this model is applicable to subcutaneous tumors, lung metastases, and intracranial tumors and offers a solution to many of the limitations of previous models. As proof of principle, we use this model to show the efficacy of a live, Listeria monocytogenes vaccine expressing the melanoma antigen tyrosinase-related protein-2 to protect mice against intravenous B16 melanoma challenge. Additionally, we extend our approach to include the human A375 melanoma model and are able to show in vivo differences between sub-lines with varying metastatic potential. These models represent an accurate and reproducible means for in vivo melanoma monitoring in preclinical studies.