Results/Conclusions Here we show that non-additive diversity effects on soil C and N cycling can be mechanistically predicted by defining plant litter inputs to soil in terms of functional chemical composition and functional chemical diversity. We found that the composition of litter chemical traits, and the ability to account for interactions among specific groups of chemical traits, was critical to predicting non-additive soil respiration (R2=0.33) and N mineralization (R2=0.52) responses to species diversity. In contrast, traditional litter chemistry metrics (e.g. %N, C:N, phenolic:N and lignin:N ratios) did not correlate well with non-additive soil C and N cycling rates (R2<0.1 for these measures of litter chemistry). We also found that effects of species richness on non-additive soil processes were likely explained by underlying changes in litter chemical diversity, and that increasing litter chemical diversity facilitates strong interactions among functional groups of compounds. These observations indicate functional chemical traits and their diversity may help explain apparently stochastic species diversity effects on non-additive decomposition processes.