Preoperative electrostimulation tests were performed on 43 postlingually deaf and 20 prelingually deaf cochlear implant (CI) candidates. The stimulating electrode was placed at three different locations, i.e. the round window, the promontory and the ear canal and the results were compared. The evoked sensations were reported to be of auditory origin by most of the postlingually deaf CI candidates. Prelingually deaf subjects could not always distinguish clearly and reliably between "hearing" and "feeling". The percentage of stimulated ears of postlingually deaf subjects in whom hearing sensations were evoked was almost identical for the three locations of the stimulating electrode. However, in 5 out of the 7 ears without hearing sensations during ear canal stimulation (ECS), hearing sensations could be evoked during either promontory stimulation (PS) or round window stimulation (RWS). The mean threshold level for "hearing" at a stimulation frequency of 62 Hz was lowest during RWS, 7.7 dB higher during PS and 35.8 dB higher during ECS. The mean electrical dynamic range at 62 Hz was most favourable during RWS (23.9 dB), smaller during PS (15.6 dB) and smallest during ECS (10.0 dB). All differences were statistically significant. Placement of the ear canal electrode was easier, less frightening for the patient and required less time than insertion of the needle electrode for PS or RWS. Therefore we recommend the use of ECS to examine whether the CI candidate can be stimulated, and of RWS if more detailed information is required.