We tested the interaction of face and voice information in identity recognition in both healthy controls and a patient (SB) who is unable to recognize faces (prosopagnosia). We asked whether bimodal information would facilitate identity recognition in patient SB. SB and controls learned the identities (face and voice) of individuals and were subsequently tested on two unimodal and one bimodal stimulus condition. SB’s poor identity recognition with faces only information was contrasted by his excellent performance with voices only information. SB’s performance was better in the bimodal conditions compared to that for visual faces alone, however, his performance was worse in the bimodal conditions compared to that for voices alone. Controls demonstrated the exact opposite pattern. For all participants, identity recognition was facilitated with ‘new’ stimuli from the participant’s dominant modality but inhibited with ‘new’ stimuli from the non-preferred modality. These findings demonstrate perceptual interference from the non-dominant modality when vision and audition are combined for identity recognition suggesting interconnectivity of the visual and auditory identity pathways. Moreover, in spite of an inability to recognize faces due to damage to face identity pathways, residual interconnectivity between voice and face processing interferes with auditory identity recognition.