Several reports have indicated that some cochlear implant patients experience a reduction in their tinnitus while listening to noise or speech. In the present study, two patients reporting bilateral tinnitus were selected from a group of adults with the Nucleus cochlear implant. They rated their tinnitus loudness and severity and completed the Tinnitus Handicap Questionnaire. The stimuli for electrical stimulation were charge-balanced pulse trains of various repetition rates (frequencies) and inter-electrode distances. A range of electrodes was chosen in each subject, including basal, medial and apical electrodes. For each condition, the hearing threshold level and the uncomfortable loudness level were determined. A range of stimulus levels between these two values were presented randomly. After each presentation, the patients rated the stimulus loudness and the tinnitus loudness on a 0 to 100 scale. These judgements were used to carefully determine the psychometric function between stimulus level and stimulus loudness, and between stimulus level and tinnitus loudness. All the parameters explored were important for maximizing the relationship between tinnitus reduction and stimulus loudness. First, the effectiveness of electrical stimulation in tinnitus reduction depended on the place along the cochlear partition. Second, a pulse rate of 125 Hz showed the greatest efficiency in terms of the current level needed to suppress tinnitus. Third, these two subjects showed rather poor performances in speech perception when using their speech processor in the usual condition and the hypothesis of an influence from tinnitus annoyance is suggested in addition to some more classical predicting factors of speech recognition in cochlear implant users.