Little is known about the role of plant functional diversity for ecosystem-level carbon (C) fluxes. To fill this knowledge gap, we translocated monoliths hosting communities with four and 16 sown species from a long-term grassland biodiversity experiment (‘The Jena Experiment’) into a controlled environment facility for ecosystem research (Ecotron). This allowed quantifying the effects of plant diversity on ecosystem C fluxes as well as three parameters of C uptake efficiency (water and nitrogen use efficiencies and apparent quantum yield). By combining data on ecosystem C fluxes with vegetation structure and functional trait-based predictors, we found that increasing plant species and functional diversity led to higher gross and net ecosystem C uptake rates. Path analyses and light response curves unravelled the diversity of leaf nitrogen concentration in the canopy as a key functional predictor of C fluxes, either directly or indirectly via LAI and aboveground biomass.