The magnitude of terrestrial carbon (C)-climate feedback largely depends on the temperature sensitivity of soil organic matter (SOM) decomposition (Q₁₀). However, our understanding of determinants of Q₁₀ for SOM fractions such as particulate and mineral-associated organic matter (POM and MAOM, respectively) is still inadequate. Particularly, it remains unclear whether microbial effects on Q₁₀ are fraction-dependent, which induces large uncertainties in projecting soil C dynamics. Here, we conducted large-scale topsoil sampling on the Tibetan Plateau, in combination with SOM fractionation and 300-day laboratory incubation to assess SOM fraction-dependent linkages between Q₁₀ and microbial properties. We found that compared with MAOM, POM had larger Q₁₀ and greater microbial diversity, and also structured distinct microbial communities as well as their co-occurrence patterns. Furthermore, associations of Q₁₀ with microbial properties differed between the two SOM fractions. Bacterial community composition and relative abundance of bacterial keystone taxa affected Q₁₀ for POM and MAOM respectively, while bacterial alpha diversity showed opposite relationships with Q₁₀ for POM and MAOM. These findings highlight the necessity of incorporating SOM fraction-dependent microbial properties and their linkages with Q₁₀ into Earth system models to accurately predict terrestrial C-climate feedback.