Background and aims We carried out field experiments to investigate if an agricultural grassland mixture comprising shallow-(perennial ryegrass: Lolium perenne L.; white clover: Trifolium repens L.) and deep-(chicory: Cichorium intybus L.; Lucerne: Medicago sativa L.) rooting grassland species has greater herbage yields than a shallowrooting two-species mixture and pure stands, if deep-rooting grassland species are superior in accessing soil 15 N from 1.2 m soil depth compared with shallow-rooting plant species and vice versa, if a mixture of deep-and shallow-rooting plant species has access to greater amounts of soil 15 N compared with a shallow-rooting binary mixture, and if leguminous plants affect herbage yield and soil 15 N-access. Methods 15 N-enriched ammonium-sulphate was placed at three different soil depths (0.4, 0.8 and 1.2 m) to determine the depth dependent soil 15 Naccess of pure stands, two-species and four-species grassland communities. Results Herbage yield and soil 15 N-access of the mixture including deep-and shallow-rooting grassland species were generally greater than the pure stands and the two-species mixture, except for herbage yield in pure stand lucerne. This positive plant diversity effect could not be explained by complementary soil 15 N-access of the different plant species from 0.4, 0.8 and 1.2 m soil depths, even though deep- Conclusions Our study showed that increased plant diversity in agricultural grasslands can have positive effects on the environment (improved N use may lead to reduced N leaching) and agricultural production (increased herbage yield). A complementary effect between legumes and non-leguminous plants and increasing plant diversity had a greater positive impact on herbage yield compared with complementary vertical soil 15 N-access.Plant Soil DOI 10.1007/s11104-013-1694-0 Responsible Editor