AVATAR therapy offers a unique therapeutic context that uses virtual reality technology to create a virtual embodiment of the voice-hearing experience, enabling the person to visualize their persecutory voice and engage in real-time "face-to-face" dialogue. The present study explores, for the first time, the contribution of sense of voice presence, together with session-by-session reduction of anxiety and paranoid attributions about the avatar, to changes in primary outcomes following AVATAR therapy. Data from 39 participants, who completed AVATAR therapy and attended a 12-week follow-up assessment, were analysed. Mid- to high-levels of sense of voice presence were reported across the therapy sessions, along with significant reductions of anxiety levels and paranoid attributions about the avatar. The interaction of sense of voice presence and reduction of anxiety was associated with two of the significant therapy outcomes: PSYRATS total and frequency of voices. The findings suggest that improvements in voice severity and frequency at post AVATAR therapy may be influenced by the combination of feeling less anxious in the context of a realistic simulation of the voice, while voice-related distress may involve additional cognitive and relational processes.