ABSTRACT Small extracellular vesicles (sEVs) are an important part of intercellular communication. They are phospholipid bilayer particles that carry active biomolecules such as proteins, various nucleic acids, and lipids. In recipient cells, sEVs can alter cellular functions, including cancer development and premetastatic niche formation in distant organs. Moreover, sEVs can carry cancer-specific features, which makes them promising biomarker candidates. However, the interactions of sEVs with biological barriers and consequences thereof, are not clarified yet. The blood-saliva barrier is crucial for preventing the entry of pathogens and (in)organic substances into the bloodstream, as well as molecule filtration from blood to saliva. The effects of brain derived DU145 prostate cancer (PCa) sEVs on a human submandibular salivary gland barrier (SSGB) in vitro were investigated. Small EVs were harvested from normoxic (N, atmospheric O2) or hypoxic (H, 1% O2) conditions, fluorescently labeled with CellTrackerTM Orange and thoroughly characterized. HTB-41 B2 cells were used as SSGB model cultured on 24-well ThinCert® inserts. After model optimization indicating effects of serum and serum-sEVs on barrier properties, PCa sEVs were applied to the basolateral (blood) side in either 10% serum, or serum-free conditions, and barrier integrity was continuously monitored for 40 hours. This study found that H and N PCa sEVs were uptaken by the SSGB in vitro model in similar quantities regardless of the media composition in the basolateral compartment. Permeation of fluorescent PCa sEVs into the apical compartment was not detectable with the applied methods. However, treatment with H and N sEVs under different serum conditions revealed distinct molecular clusters after hierarchical analysis of mRNA data measured by high-throughput qPCR, which were partly reflected at the protein level. For example, serum-reduction dependent decrease of barrier properties was accompanied with the decrease of CDH1 or Claudin-7 expression. Interestingly, the presence of H sEVs significantly increased the number of sEV-sized particles in the apical compartment of the SSGB model compared to basolaterally added N sEVs. This functional effect on the number of particles in the saliva (apical) compartment induced by different sEVs applied in the blood (basolateral) compartment might be a new approach to understand one possible mechanism how differences of salivary EVs might occur which then could be used as biomarker.