Subtropical forests play an important role in regulating global carbon storage, a substantial portion of these forest carbon pools are stored in biomass stocks. Yet, we do not fully understand how tree diversity, stand structure and functional trait identity influence biomass stocks along elevational gradients. Here we used forest inventory data from nine 1-ha plots across different elevational gradients in subtropical forests of southern China. We analyzed the effects of tree diversity, structure and functional trait identity on biomass along elevational gradients, and tested the complementarity effect, selection effect, mass-ratio hypothesis and specie-energy hypothesis. We found that multiple metrics of diversity and structural attributes significantly promoted biomass. Forest biomass was improved by traits with greater maximum height, larger seed mass (SM) and lower wood density (WD). Specifically, large-diameter trees were the strongest independent predictor of biomass relative to other single variables. A significant increase in tree diversity, density and biomass was observed with increasing elevational gradients; additionally, trees dominated by traits related to greater maximum tree height, larger SM and lower WD also increased. Elevation indirectly increased biomass by increasing diversity and community-weighted mean traits, but indirectly decreased it by reducing structural attributes. Our results support the complementarity effect, selection effect and mass-ratio hypothesis simultaneously, but not the species-energy hypothesis. The selection effect had greater effects on stand biomass than the complementarity effect and mass-ratio hypothesis. Our findings suggest that developing natural tree diversity and structural complexity as well as dominant resource-acquisitive species is beneficial to maintain and enhance carbon storage capacity in species-rich subtropical forests along elevational gradients.