Objective: To seek evidence for perilymph- cerebrospinal fluid barrier (CSF). Methods: A well-controlled comprehensive comparison of kinetics of hydrocortisone (HC) and methylprednisolone (MP) in perilymph of scala tympani (PST), perilymph of scala vestibuli (PSV), scala media endolymph (SME), CSF and plasma was made in guinea pigs. Results: One hour post intratympanic injection, the mean HC concentrations were exactly the same for both the PST and PSV. The concentrations in the two cochlear partions rapidly decreased from 1 h to 2 h post injection. Thereafter it decreased gradually. HC concentration was undetectable in CSF following direct intratympanic application. Likewise, topical application of MP resulted in the rapid appearance of the substance in both the PST and PSV. The mean value of MP concentration in PSV were similar to those in PSV. There was no significant difference in MP concentrations between the PST and PSV although there appeared to be a trend at 1, 2 and 6 h for concentrations to be greater in PSV than in PST. No detectable concentration of MP in CSF was observed after intratympanic administration. Both the HC and MP levels were significantly higher in SME than in the two perilymphatic compartments throughout the experiment. Conclusion: The findings of the experiment indicate that there is no open communication via the cochlear aqueduct that would allow PST to enter CSF. The CSF cannot directly enter the scala tympani via the cochlear aqueduct, displace and ultimately replace PST. Accordingly, the present study confirms the presence of a tight perilymph-CSF barrier.