UNLABELLED: Cochlear implants may improve the medical and social prognosis of profound deafness. Nevertheless, some patients have experienced poor results without any clear explanations. One correlate may be an alteration in cortical voice processing. To test this hypothesis, we studied the activation of human temporal voice areas (TVA) using a well-standardized PET paradigm adapted from previous functional MRI (fMRI) studies. METHODS: A PET H(2)(15)O activation study was performed on 3 groups of adult volunteers: normal-hearing control subjects (n = 6) and cochlear-implanted postlingually deaf patients with >2 y of cochlear implant experience, with intelligibility scores in the "Lafon monosyllabic task" >80% (GOOD group; n = 6) or <20% (POOR group; n = 6). Relative cerebral blood flow was measured in 3 conditions: rest, passive listening to human voice, and nonvoice stimuli. RESULTS: Compared with silence, the activations induced by nonvoice stimuli were bilaterally located in the superior temporal regions in all groups. However these activations were significantly and similarly reduced in both cochlear implant groups, whereas control subjects showed supplementary activations. Compared with nonvoice, the voice stimuli induced bilateral activation of the TVA along the superior temporal sulcus (STS) in both the control and the GOOD groups. In contrast, these activations were not detected in the POOR group, which showed only left unilateral middle STS activation. CONCLUSION: These results suggest that PET is an adequate method to explore cochlear implant benefits and that this benefit could be linked to the activation of the TVA.