We investigated how soil fauna impact leaf litter decomposition in north temperate forests using litter bags with different mesh sizes (5 mm, 0.5 mm, and 0.01 mm) to exclude specific fauna by size. The experimental design included two regions (warmer, cooler) in New York State, two forest types (coniferous, deciduous), and litter bags with four tree species (yellow birch, sugar maple, red oak, red pine) that varied in litter resource quality. Excluding most soil fauna with 0.01-mm mesh decreased mass loss from litter bags by 8.8 % overall; by 19 % in the relatively warmer region; by 12 % in coniferous forest stands; and by 18 % for high-quality yellow birch litter but not for high-quality sugar maple litter or low-quality red oak and red pine litters. Mass loss rates were predicted by initial nitrogen concentrations and by initial ratios of carbon / nitrogen and lignin / nitrogen of the leaf litter; these relationships were stronger for litter bags that excluded soil fauna. Fauna extracted from the leaf litter residue were predominantly Acari (mites) and Collembola (springtails). Soil fauna mediated the extent that biochemical fractions (nitrogen, hemicellulose, cellulose, lignin) were lost from decomposing litter but in idiosyncratic ways. Not only are soil fauna impacts on leaf litter decomposition widespread, but, as shown here, they can be idiosyncratic when evaluated with several litter species that differ in resource quality, placed in different forest types, and across a climate gradient. Future analyses of soil fauna impacts on leaf litter decomposition should combine litter-, site-, and climate-related variability to improve understanding and enable prediction.