Brain metastases are clinically diagnosed in the majority of patients with metastatic melanoma. The prognosis for patients with melanoma brain metastasis is poor with a median survival time of 6 months after diagnosis. Development of better therapies requires a better understanding of the biology of melanoma brain metastasis. The development of a relevant in vivo model offers this possibility. The intracarotid injection of different murine or human melanoma cells into syngeneic or nude mice produces metastases in different regions of the brain. This site-specific metastasis is not due to patterns of initial cell arrest, motility, or invasiveness, but rather to the ability of melanoma cells to proliferate in the brain parenchyma or the meninges. The blood-brain barrier is intact in metastases that are smaller than 0.25 mm in diameter. Although in larger metastases the blood-brain barrier is leaky, the lesions are resistant to many chemotherapeutic drugs. We have also analyzed the malignant behavior of several melanoma cell lines isolated from brain or visceral metastases of patients. The cells from brain metastases showed a slower growth rate and exhibited lower metastatic potential than cells from visceral metastases, indicating that brain metastases do not necessarily represent the end stage in the metastatic cascade. Rather, brain metastases are likely to originate from a unique subpopulation of cells within the primary neoplasm.