Patients undergoing cochlear implantation could benefit from a simultaneous application of drugs into the ear, helping preserve residual low-frequency hearing and afferent nerve fiber populations. One way to apply drugs is to incorporate a cannula into the implant, through which drug solution is driven. For such an approach, perilymph concentrations achieved and the distribution in the ear over time have not previously been documented. We used FITC-labeled dextran as a marker, delivering it into perilymph of Guinea pigs at 10 or 100 nL/ min though a cannula incorporated into a cochlear implant with the outlet in the mid basal turn. After injections of varying duration (2 hours, 1 day or 7 days) perilymph was collected from the cochlear apex using a sequential sampling technique, allowing dextran levels and gradients along scala tympani to be quantified. Data were interpreted quantitatively using computer simulations of the experiments. For injections of 2 hours duration, dextran levels were critically influenced by the presence or absence of fluid leakage at the cochleostomy site. When the cochleostomy was fluid-tight, substantially higher perilymph levels were achieved at the injection site, with concentration declining along scala tympani towards the apex. Contrary to expectations, large dextran gradients along scala tympani persisted after 24 hours of sustained injection and were still present in some animals after 7 days injection. Functional changes associated with implantation and dextran delivery, and the histological state of the implant and cannula were also documented. The persistent longitudinal gradients of dextan along the ear were not readily explained by computer simulations of the experiments based on prior pharmacokinetic data. One explanation is that inner ear pharmacokinetics are altered in the period after cochlear implantation, possibly by a permeabili-zation of the blood-labyrinth barrier as part of the immune response to the implant.