Biodiversity research has undergone major changes over the past decade and a half, with a boom in experiments investigating the link between the diversity of organisms and ecosystem processes and functioning. The main approach in such experiments is to create a gradient of plant diversity, by randomly selecting species for the mixtures from a species pool. As a consequence, although the species pool is based on natural assemblages of plant species from a given habitat, the actual assemblages created in such an approach usually have no exact natural correlate. In essence one deconstructs and then reconstructs nature. In restoration projects on the other hand, humans change abiotic conditions and try to establish desired species in degraded ecosystems - they reconstruct nature. This talk explores the vast potential for cross fertilization between these two fields of biodiversity and restoration research. Biodiversity experiments can tell us how well a system is able to provide certain ecosystem services such as nutrient cycling or decomposition). In addition they can provide an idea of a minimum number of necessary species for a system to be able to function, given certain environmental conditions. All ecosystems in a modern landscape can be seen as biogeochemical systems somewhere along a continuum: from a balanced relationship between biodiversity and ecosystem functioning to a malfunctioning relationship. One main goal of restoration could therefore be to restore the system back to a functioning relationship between biodiversity and ecosystem functioning. Now we need to assess how applicable the results from biodiversity experiments are under more natural also under more extreme conditions, such as those often encountered in restoration projects. I discuss new experiments testing positive biodiversity effects as found in experiments, in controlled as well as restoration-relevant settings.