Inspired by grassland biodiversity experiments studying the impact of plant diversity on primary productivity, the Crop Diversity Experiment setup in 2018 aimed at testing whether these biodiversity benefits also hold for annual crop systems and whether crop mixtures also achieved transgressive overyielding, i.e. yield in mixture that was higher than the most productive monoculture. The first 3 years of the experiment demonstrated that crop mixtures do not only increase yield compared with an average monoculture but often also compared with the highest yielding monoculture. The crop diversity effects were stronger under more stressful environmental conditions and were often achieved in mixtures with legume crops. However, we observed transgressive overyielding also under favorable conditions and in mixtures without legumes. With our investigation of the underlying mechanisms of the yield benefits we found both direct complementarities between crop species and indirect effects via other organisms. The former included chemical, spatial and temporal complementarity in Nuptake, complementary root distribution leading to complementary water uptake, as well as spatial and temporal complementarity in light use. Among the indirect mechanisms we identified complementary suppression of weeds and more abundant plant growth-promoting microbes in crop mixtures, apart from complementarity in pest and disease suppression not yet studied in the Crop Diversity Experiment but demonstrated elsewhere. In consequence, the Crop Diversity Experiment supports not only the assumption that the ecological processes identified in biodiversity experiments also hold in crop systems, but that diversification of arable crop systems provides a valuable tool to sustainably produce food.