The Mo stable isotope system has been used to trace material recycling during subduction-related processes, but the behavior of Mo isotopes during magmatic evolution (e.g., crystal–melt fractionation and melt–fluid interaction) remains contentious, especially in high-SiO₂ granites. This study addresses the issue of Mo isotope variation in high-SiO₂ granites by measuring bulk-rock and mineral Mo isotopes of biotite granites (BGs) and garnet-bearing two-mica granites (GBGs) from the well-characterized Zhengga granite pluton (southern Tibet, China). The GBGs have similar Sr–Nd–O isotope compositions to those of the BGs but show higher SiO₂ and lower TiO₂, MgO, total Fe₂O₃, and CaO contents, and represent the products of advanced fractionation of the BG magmas. The BGs have lower Mo contents (0.02–0.07 ppm) and higher δ^98/95Mo values (−0.54‰ to 0.22‰) compared with the GBGs (0.029–2.121 ppm and −0.97‰ to −0.41‰, respectively). Analysis of major silicate minerals suggests that substantial segregation of biotite and feldspar with high δ^98/95Mo values of 0.00‰ to 0.38‰ and −1.06‰ to 0.57‰ (most within −0.58‰ to 0.13‰) could have driven the GBGs and the late-stage crystalline phase of garnet (−1.22‰ to −0.98‰) towards very low δ^98/95Mo values. However, the trend of decreasing δ^98/95Mo with indices of magma differentiation is not linear: one group of GBGs show increasing Mo contents and decreasing δ^98/95Mo values with decreasing Y, Ho, and Dy contents; while the other group display increasing Mo contents and slightly decreasing δ^98/95Mo values with respect to the increasing contents of Y, Ho, and Dy. These two contrasting behaviors can be ascribed to further crystal fractionation and melt–fluid interaction in a closed magmatic–hydrothermal system. This is also evidenced by the formation of two types of garnets with different contents of Mo and rare earth elements in these two groups of GBGs. Closed-system fluid saturation is inferred to have driven the silicate melt to be enriched in ⁹⁸Mo, which limited the decrease in melt δ^98/95Mo caused by crystal fractionation. These observations are supported by quantitative geochemical modeling. We conclude that both fractional crystallization and melt–fluid interaction control Mo isotope fractionation in high-SiO₂ granites and that Mo isotopes are useful for tracing the evolution of high-SiO₂ igneous rocks.