For 20 years direct electrical stimulation has been used in cases of severe bilateral hearing loss or complete deafness to mediate acoustic percepts. The relevant literature is reviewed. While these attempts were initially thought to be unphysiological and unsuited for transferring speech, acoustic sensations were successfully conveyed even with the very simple unichannel electrodes through percutaneous signal transfer. Patients fitted with such simple systems were able to hear and distinguish environmental noise and speech, their lip reading as well as their speech improved. Speech discrimination was, however, impossible with such simple implants both on theoretical grounds and in practical terms, because frequency analysis is exclusively based on periodicity (up to 400 Hz). Designing bipolar multichannel electrodes which, when introduced into the scala tympani or the modiolus, produce discrete stimulation of several circumscribed groups of nerve fibers, was the logical consequence of earlier attempts along these lines, Implantation of these systems can be done along the transmeatal, meato-mastoidal or mastoidal approach. The electrodes can be implanted in bundles through the round window or into the modiolus; they can, however, also be introduced individually through several drill holes in the promontory for placement in the scala tympani and vestibuli. This produces a far more differentiated stimulation simulating a tonotopic pattern of stimuli. In addition to periodicity, the place principle can thus be utilized for frequency coding. While their dynamic range is rather poor (15 to 30 dB at best), multichannel systems, in theory, offer substantially more favorable conditions for speech intelligibility. Since current knowledge of speech coding is, however, inadequate, the degree of intelligibility obtainable is still insufficient for everyday life. Inspite of this flaw, such implants as are available today substantially benefit the patients, who are able to establish acoustic communications with their environment by distinguishing environmental noise from speech, to discriminate between male and female voices, to recognize musical rhythms and even to understand a few words. Indications for the implantation of prostheses, the requisite conditions and postoperative training programs are discussed. While cochlear implants are still experimental, they appear to be reasonably justified in selected cases, since they have been well tolerated by all patients treated sofar without causing any complications and since many of the data obtainable can only be collected in humans. It is, however, essential that experimental implantation be exclusively dealt with by specialized teams, which should evaluate such data as are available and translate them into practice as soon as possible. A routine impantation of hearing prostheses is currently unwarranted.