Partial melts derived from crustal anatexis exhibit large variations in field characteristics, composition, and isotope geochemistry. The relative influence of melting process and magmatic differentiation on such heterogeneity is a subject of ongoing debate. Boron and molybdeum isotopes have the potential to shed light on the genesis of crustal melts, but the behaviors of B and Mo isotopes during crustal anatexis and differentiation are unclear. Here, we present geochemical and isotopic data, including whole-rock Sr–Nd–B–Mo and mineral B–Mo data, for the Conadong leucogranites in the Himalayan Block, to provide new insights into these issues. The studied samples include two-mica granites (biotite + muscovite), muscovite leucogranites (muscovite ± garnet), and biotite-rich granites. The two-mica granites were derived from muscovite dehydration melting of metasedimentary rocks. Their initial Sr and Nd isotopic compositions show positive correlation, implying non-modal partial melting associated with dissolution of accessory phases. The B contents (3.44–21.2 ppm) of the two-mica granites decrease but their δ¹¹B (−15.80‰ to −13.25‰) and δ⁹⁸Mo (−0.15‰ to 0.34‰) values increase with increasing Sr and Ba contents. This may have been controlled by different phases involved in non-modal melting (i.e., muscovite and feldspar). The muscovite leucogranites display features of residual liquids. Their δ¹¹B (−15.38‰ to −11.90‰) and δ⁹⁸Mo (−0.15‰ to 0.17‰) values negatively correlate with Sr and Ba contents, which can be ascribed to shallow-crustal crystal–liquid separation and melt–fluid interaction. The biotite-rich granites are least evolved, with large variations in their initial Sr–Nd isotopic compositions. They show relatively heavy whole-rock B isotopic compositions of −11.97‰ to −9.00‰. These may indicate that biotite-rich granites may be contaminated by entrained restitic minerals (e.g., biotite, feldspar and accessory minerals). Overall, this study demonstrates that compositional diversity of anatectic melts is controlled primarily by melting processes in the source and secondarily by emplacement-level crystal–liquid separation. Both non-modal partial melting and crystal–liquid separation could result in B–Mo isotopic fractionation in anatectic melts.