Metastasis from the primary tumor site to the brain is the most lethal complication of advanced breast cancer. There is no translational approach to detect if a primary tumor has brain metastatic potential. This is due to a lack of blood brain barrier (BBB) models that can classify a cells metastatic potential. Moreover, the mechanisms by which circulating cancer cells extravasate through the BBB are unknown. Currently used in vivo murine models are slow to manifest metastasis therefore, an alternative approach is an in vitro microfluidic model that re-capitulates the BBB niche micro-environment. Our goal is to develop and use this model to identify the metastatic potential of cancer cell populations. Therefore, we have developed a blood brain niche (µm-BBN) on-a-chip and studied the phenotypic differences between cancer cells in the µm-BBN. The device has two chambers separated by a 5µm porous membrane coated with Matrigel and a BB human endothelial cell (hCMEC/D3) mono-layer. The bottom chamber contains a brain stromal ECM and Normal Human Astrocytes (NHA), while the top chamber acts as the blood vessel. Cancer cells are introduced into the top chamber and allowed to extravasate into the brain like stroma. We measured the ability of the endothelial layer to prevent fluorescent small molecules from diffusing into the brain stromal space. The barrier measured 8.3X lower max fluorescent values than when no barrier was present and were confirmed by TEER. Using this model, we characterized the MDA-231 breast cancer cell line against a brain-seeking subclone (MDA-231-BR), normal-like cell lines (MCF10A) and brain met patient derived xenografts (PDXs) in terms of their ability to extravasate, migrate and survive in the niche for 24-48 hrs. Phenotypic and migratory behavior was recorded using confocal tomography to measure cancer cell properties (volume, shape, position) relative to the endothelial layer. Brain-seeking subclones cluster around the endothelial layer, the MCF10A cell line has no preferred location and the parent line (MDA-231) extravasates deeper into the brain stromal space than the other two cell lines. We also found significant variation in the shape of each cell line before and during extravasation suggesting differences in plasticity. The effects of chemoattractants within the µm-BBN on extravasation have been explored by omitting astrocytes in the collagen-infused brain niche and replacing with astrocyte conditioned media. These findings confirm that the system is capable of measuring both variations in cancer cell populations and individual cells. This approach may enable classification of subclone populations with higher metastatic potential, meeting a major need in Oncology. Future work will employ this emerging tool to study the mechanisms by which the cancer cells extravasate and survive in the niche.