【Objective】Species abundance distribution is a classic cutting-edge research field in the study on community ecology. Niche apportionment models are commonly used to simulate relative abundance distribution of species, but quite rarely applied to the study of soil fauna. Moreover, little has been reported in the past on how soil fauna responds to environmental changes in community construction relative to feeding guild. A field investigation was carried out of relative species abundance (individuals and biomass) distribution of the litter-dwelling fauna as a whole as well as each feeding guild (including omnivores, phytophage, predators, and saprovores) in a Quercus wutaishansea forest with an elevational gradient in the Dongling Mountain, Beijing. 【Method】Five commonly used niche apportionment models, that is the broken stick model (BSM), niche preemption model (NPM), niche overlapping model (ONM), random assortment model (RAM), and dominance preemption model (DPM), were used separately to simulate abundance distribution of the fauna. Adequacy of sampling is the prerequisite for the models to simulate species relative abundance distribution and judged by species accumulation curves. Rank-abundance plots were used to show richness and evenness of the species contained in the litter-dwelling soil fauna and each feeding guild. 【Result】 Results show that sampling in this study was adequate for fitting of the soil fauna community and feeding guilds. The distribution of the soil fauna in individual number disagreed with that in biomass. The dominance preemption model was not fit for either the whole litter-dwelling fauna community or each feeding guild based on chi-square test. The broken stick model was the optimal one for fitting the whole litter-dwelling fauna community, free of any impact of changes in elevation. The overlapping model and the niche preemption model were also good enough in fitting. When individual number was used as the index for abundance, elevation did not affect much fitting of the models for each feeding guild of the litter-dwelling soil fauna according to the chi-square test, but the chi-square test did vary from guild to guild. For omnivores, only the dominance preemption model failed the chi-square test, but for phytophage and predators, the random assortment model and for saprovores, the niche preemption model did, too. When biomass was used as the index for abundance, elevation did not have any impact on the chi-square test of the models fitting omnivores and predators, but did on that of the dominance preemption model fitting omnivores, and of the dominance preemption model and the niche preemption model fitting predators. The chi-square test of the models fitting saprovores and phytophage varied with the elevation. For saprovores, the dominance preemption model was good in the section low or moderate in elevation, while in the seciton high in elevation, the dominance preemption model and the random assortment model were. For phytophage, the dominance preemption model was good in the section low in elevation, the dominance preemption model and the random assortment model were in the section moderate in elevation, but the dominance preemption model, the niche preemption model and the random assortment model all failed the chi-square test in the section high in elevation. Abundance distribution of the various feeding guilds of the fauna was very sensitive to changes in elevation. For omnivores, the optimal model changed from the niche preemption model to the broken stick model with rising elevation, while for phytophage, it did reversely. Fitting with individual number as the index of abundance indicates that the optimal model for predators changed from the niche preemption model to the broken stick model and the niche preemption with rising elevation. For saprovores, the optimal model was the random assortment model in the sections low and high in elevation, while it was the broken stick model in the section moderate in elevation. When biomass was used as the index of abundance, the broken stick model replaced the random assortment model as the optimal one for predators, and the niche preemption model for saprovores. The use of biomass was better than that of individual number as index for the models to fit abundances of various feeding guilds of soil fauna. 【Conclusion】To sum up, elevation and selection of index does not affect much the models in fitting abundance distribution of the soil fauna as a whole, but does in fitting abundance distribution of the various feeding groups of the soil fauna. Future studies should consider modern theory of species coexistence and the neutral theory to examine the relative importance of stabilizing mechanism and equalizing mechanism in community assembly.