Terrestrial and aquatic ecosystems contain a large number of species and provide a large number of ecosystem services, in particular transformation of solar energy and carbon dioxide into biomass, fueling element cycles, soil formation, climate control and more. The role of biodiversity–ecosystem functioning experiments is to test what would happen if the number of species within an ecosystem would be reduced—more drastically than experienced up to now but predicted under some scenarios for the future. These experiments have shown that species loss strongly reduces ecosystem functioning and ecosystem stability and that this effect increases over time. The well-functioning and stability of species-rich ecosystems seem to mainly be due to reduced competition and beneficial interactions between species with complementary niches and/or at different trophic levels. However, it is still unclear what makes interactions beneficial (e.g., which trait differences), how such interactions are shaped by community-assembly and evolutionary processes, and which forces may counteract the emergence of positive biodiversity effects in ecosystems. The challenge for future experiments is to find out new designs that could address these open questions. More direct manipulations of specific interactions, either in simplified model systems or by removal in more complicated systems, may be one approach. Incorporating genetic analysis underpinning interactions from the within-species up to the between-trophic-group level promises to overcome limitations of trait-based approaches. It may be particularly rewarding to develop biodiverse agricultural and forestry systems with desirable functioning, because in these cases some of the forces counteracting positive biodiversity effects such as invasion by “selfish” genotypes or species can be prevented.