Positive species richness-productivity relationships are common in biodiversity experiments, but how resource availability modifies biodiversity effects in grass-legume mixtures composed of highly productive species is yet to be explicitly tested. We addressed this question by choosing two grasses (Arrhenatherum elatius and Dactylis glomerata) and two legumes (Medicago × varia and Onobrychis viciifolia) which are highly productive in monocultures and dominant in mixtures (the Jena Experiment). We established monocultures, all possible two- and three-species mixtures, and the four-species mixture under three different resource supply conditions (control, fertilization, and shading). Compared to the control, community biomass production decreased under shading (-56 %) and increased under fertilization (+12 %). Net diversity effects (i.e., mixture minus mean monoculture biomass) were positive in the control and under shading (on average +15 and +72 %, respectively) and negative under fertilization (-10 %). Positive complementarity effects in the control suggested resource partitioning and facilitation of growth through symbiotic N2 fixation by legumes. Positive complementarity effects under shading indicated that resource partitioning is also possible when growth is carbon-limited. Negative complementarity effects under fertilization suggested that external nutrient supply depressed facilitative grass-legume interactions due to increased competition for light. Selection effects, which quantify the dominance of species with particularly high monoculture biomasses in the mixture, were generally small compared to complementarity effects, and indicated that these species had comparable competitive strengths in the mixture. Our study shows that resource availability has a strong impact on the occurrence of positive diversity effects among tall and highly productive grass and legume species.