The aim of this study was to better understand the long term behavior of silicone-based cochlear implants loaded with dexamethasone: in vitro as well as in vivo (gerbils). This type of local controlled drug delivery systems offers an interesting potential for the treatment of hearing loss. Because very long release periods are targeted (several years/decades), product optimization is highly challenging. Up to now, only little is known on the long term behavior of these systems, including their drug release patterns as well as potential swelling or shrinking upon exposure to aqueous media or living tissue. Different types of cylindrical, cochlear implants were prepared by injection molding, varying their dimensions (being suitable for use in humans or gerbils) and initial drug loading (0, 1 or 10%). Dexamethasone release was monitored in vitro upon exposure to artificial perilymph at 37 °C for >3 years. Optical microscopy, X-ray diffraction and Raman imaging were used to characterize the implants before and after exposure to the release medium in vitro, as well as after 2 years implantation in gerbils. Importantly, in all cases dexamethasone release was reliably controlled during the observation periods. Diffusional mass transport and limited drug solubility effects within the silicone matrices seem to play a major role. Initially, the dexamethasone is homogeneously distributed throughout the polymeric matrices in the form of tiny crystals. Upon exposure to aqueous media or living tissue, limited amounts of water penetrate into the implant, dissolve the drug, which subsequently diffuses out. Surface-near regions are depleted first, resulting in an increase in the apparent drug diffusivity with time. No evidence for noteworthy implant swelling or shrinkage was observed in vitro, nor in vivo. A simplified mathematical model can be used to facilitate drug product optimization, allowing the prediction of the resulting drug release rates during decades as a function of the implant's design.