Recent biodiversity–ecosystem functioning experiments in temperate grasslands have shown that productivity positively correlates with plant species richness. Resource partitioning (in particular, nitrogen [N] partitioning) has been proposed as one possible mechanism to explain this pattern. There is evidence for interspecific differences in chemical form, soil depth, and timing of N uptake. However, it has rarely been tested whether such differences result in increased N exploitation at the plant community level. Using 15N-labeled litter that was mixed into different soil layers, we tested whether eight common grasses and forbs grown in communities of one, two, or four species differ with respect to the proportions of N taken up from different soil depths (N niche), and how this affects the total N uptake of plant communities. We calculated proportional similarity between species (niche overlap) with regard to N uptake from the labeled soil layers; we further calculated an a priori measure of community N uptake based on species N uptake in monoculture (community niche). Interestingly, however, plant community N uptake was not affected by species richness, possibly because community-level N uptake was determined by (diversity-independent) soil N mineralization rates. We nevertheless observed a positive effect of species richness on productivity due to increased aboveground biomass : N ratios. This may indicate increased competition for light, resulting in increased amounts of comparably N-poor stem tissue. However, community N content and biomass were positively correlated with the community niche, a measure which is strongly linked to species composition. Thus, our results suggest that the studied species are generalists rather than specialists regarding N uptake depth, and that species composition was more important than species richness in determining community N uptake. Overall, N partitioning may be a less important driver of positive biodiversity–productivity effects in temperate grasslands than previously assumed.