1. Herbivore grazing has major effects on soil nutrient dynamics in a variety of grassland ecosystems. Previous studies have examined how large herbivores as a group affect nutrient cycling, but little information is available on how assemblages of different herbivore species may influence nutrient cycling, and whether herbivore assemblage effects are influenced by plant community characteristics (e.g., composition, diversity) of the grazed grassland. 2. We conducted a five-year, replicated grazing experiment to test the effects of different large herbivore assemblages (cattle grazing, sheep grazing, combined cattle and sheep grazing, no grazing) under moderate grazing intensity on soil nitrogen (N) mineralization rate in two types of grassland communities (high forbs/high diversity and low forbs/low diversity) in meadow steppe habitat of northeast China. Moreover, we examined two distinctly different pathways that herbivores could influence soil N availability: directly through urine and dung deposition and indirectly by shifting grassland species composition (i.e., the grass: forb ratio), thereby the quality of plant litter available to soil decomposers. 3. We found that grazer effects on soil N availability (indexed with anion and cation adsorption strips) depended on herbivore assemblage, and the herbivore assemblage effects varied in the two types of grasslands. In one type of grassland characterized by low diversity, grazing by each of the herbivore assemblages enhanced soil N availability compared to the ungrazed plots, and mixed species (cattle and sheep) grazing had a greater effect than single species grazing. In high diversity grassland, single species herbivore grazing significantly increased soil N availability, but mixed grazing had no effect. 4. Mixed linear modelling revealed that soil N availability was facilitated primarily by excreta additions to the soil and secondarily by the abundance of grasses. 5. Synthesis. Grazers increased soil N availability directly by adding readily accessible N in urine and dung to the soil. Herbivores indirectly influenced soil N availability by altering the plant composition (grass: forb cover). Both mechanisms contributed to the variation in how different herbivore assemblages affected soil N availability in the two grassland types.