To prevent the perilymph (guinea pig) from contamination with CSF during the sampling the aqueductus cochleae (AC) was blocked by injection of tissue adhesive into the meningeal aperture. The control of an exact blockage of AC was carriedout by examination of perilymph-outflow after opening the cochlea (injection of fluorescein-Na into the CSF-space), analysis of perilymph-protein-concentration, macroscopic and microscopic examination of the temporal bones. In all cochleae we have found the same morphological structures, notwithstanding whether the AC was blocked (for a time from 30 min to 7 weeks) or not: The cochlear aqueduct is filled with a mesh of mesenchymal tissue, which grows more dense towards the cochlear aperture andcontinues into the round window membrane. From scala tympani the AC is always limited by one layer of cells forming a sort of membrane (under light microscope). It seems possible that CSF moves in the inner of the round window membrane between AC and subepithelian space of middle ear mucosa, whereas perilymph of scala tympani is not in direct contact with the flow of CSF. The scala tympanic side of the round window membrane may be a big area for diffusion and there also may be an exchange between CSF and perilymph. The outflow of CSF into the cochlea after experimental opening of the cochlea is an artifact, caused by damage of pressure equilibration between CSF-space and cochlea. 30 min and 5--7 weeks after blockage no morphologicaland electrophysiological alterations from those of the control ears were to be seen. The protein concentration, however, increased significantly 5--7 weeks after blockage from normally about 200 mg/100 ml toalmost the double especially in the scala tympani (see Table 1).