Leukocyte extravasation is a key component of the innate immune response, while circulating tumor cell extravasation is a critical step in metastasis formation. Despite their importance, the mechanisms underlying leukocyte and tumor cell extravasation remain incompletely understood. Here, we developed an imaging pipeline that integrates fast, label-free live-cell imaging with deep learning-based image analysis to quantitatively track and compare the initial steps of extravasation—cell landing and arrest on an endothelial monolayer—under physiological flow conditions. We find that pancreatic ductal adenocarcinoma (PDAC) cells exhibit variable adhesion strength and flow sensitivity. Remarkably, some PDAC cells demonstrate comparable endothelial engagement as leukocytes, preferentially arresting at endothelial junctions, potentially due to increased stiffness at these sites, which leads to exposure to the underlying basal extracellular matrix. PDAC cells also tend to cluster in regions with high, heterogeneous expression of the endothelial CD44 receptor. Simulations suggest that clustering results from the combination of CD44-mediated attachment and localized flow disturbances that facilitate subsequent cell attachment. Targeting CD44 using siRNA or function-blocking antibodies, or degrading its ligand hyaluronic acid (HA) almost completely abolishes PDAC cell attachment. Overall, we demonstrate that cancer and immune cells share both common and unique features in endothelial adhesion under flow, and identify CD44 and HA as key mediators of PDAC cell arrest.