Self-perception of voice has a significant psychosocial impact on transgender persons. Research about the evolution of self-perception of voice during cross-sex hormone therapy (CSHT) is lacking. The aim of this study was to examine if self-perception of voice changes during CSHT, and if a change of serum testosterone levels as a result of CSHT can predict a change of self-perception of voice. Prospective longitudinal study. The Transsexual Voice Questionnaire (TVQ), consisting of three factors-anxiety and avoidance (AA), gender identity (GI), and voice quality (VQ)-was used. Transgender persons completed the TVQ at baseline (80 trans men and 103 trans women), after 3 and 12 months of CSHT follow-up. Trans men: From 0 to 3 months, 0 to 12 months, and 3 to 2 months of CSHT, the AA and GI scores improved. From 0 to 3 months of CSHT, the increasing testosterone level was predictive for the improvements of AA and GI scores. Trans women: From 0 to 3 months, the GI score improved. From 0 to 12 months, the AA, GI, and VQ scores improved. Improvements of self-perception of voice could not be predicted by changing serum testosterone levels. During CSHT, self-perception of voice improves in both trans men and trans women. In trans men only, the improving self-perception of voice during the first 3 months can be attributed to the CSHT. For trans women, this study supports that testosterone has acted irreversibly virializing to the voice before CSHT, if they already went through male puberty. 4. Laryngoscope, 127:2796-2804, 2017.