Binary- and multi-species sown mixtures may increase herbage yield and/or reduce inorganic nitrogen (N) requirement compared to perennial ryegrass (PRG) (Lolium perenne L.) swards. A split-plot design was used to compare yields of binary- and multi-species mixtures to single-species swards of three grasses and red clover managed for intensive silage production under varying N application rates. Perennial and Italian (Lolium multiflorum Lam.) ryegrasses had greater annual yields when grown as single species receiving 360 kg N/ha per year than in binary mixtures with red clover (Trifolium pratense L.) receiving 0 kg N/ha per year, whereas timothy (Phleum pratense L.) produced equally high yields in both situations. When no inorganic N was applied, the annual dry matter yield of Mix 1 (10,738 kg/ha; PRG, timothy, red clover and white clover (Trifolium repens L.) and Mix 2 (11,679 kg/ha; PRG, timothy, red clover, ribwort plantain (Plantago lanceolata L.) and chicory (Cichorium intybus L.)) was greater than that of a PRG sward (PRG/0N; 5,885 kg/ha) and derived more from the contribution of legumes than herbs. This yield advantage of mixtures declined as inorganic N input increased, as did the legume and herb proportions in the multi-species swards. When averaged across rates of inorganic N input, Mix 2 had a greater annual yield than Mix 1 (12,464 vs. 11,893 kg/ha). Mix 2 receiving no inorganic fertiliser N and both Mix 1 and Mix 2 receiving 120 kg N/ha per year matched the annual yield achieved by PRG receiving 360 kg N/ha per year. Our results indicate that the yield performance of binary- and multi-species grassland swards should be measured in situ rather than predicted from single-species swards of constituent species.