Extracellular vesicles (EVs) have emerged as promising biomarkers for monitoring physiological homeostasis and pathological progression. However, current analytic methods face limitations in preserving spatial information about EVs and their intricate connections to parental and recipient cells. Here, we present Spatial-EV-seq, a method for in situ spatial profiling of EVs within their native microenvironmental context. Spatial-EV-seq uses an antibody-engineered capture interface to preserve EVs’ spatial distribution, followed by rolling circle amplification with EV surface–binding aptamers enabling fluorescence imaging and molecular profiling of individual EVs. The method integrates ultra-sensitive EV profiling, molecular subtyping, and high-resolution spatial mapping with transcriptomics to resolve location-specific EV-cell communication networks. In an anti-PD-1 treated breast cancer mouse model, we uncover a spatially orchestrated immunosuppressive axis: PD-L1⁺ EV-enriched zones drive CD8⁺ T cell dysfunction, establishing immune-privileged niches, whereas PD-L1⁺ EV-depleted regions preserve immunocompetence and therapeutic sensitivity. Spatial-EV-seq offers insights into EV-mediated mechanisms and unlocks avenues for precision diagnostics and therapeutics.