Three grazing experiments were carried out in late spring (early lactation), summer (mid‐lactation) and autumn (late lactation) to compare the effects of perennial ryegrass cultivar or grass species, sown in binary or multispecies mixtures, on milk yield and nitrogen excretion of dairy cows. Replicated groups of multiparous Holstein Friesian × Jersey cows were offered either a control or high‐sugar perennial ryegrass (Lolium perenne) or tall fescue (Festuca arundinacea) base grass in a binary mixture with white clover (Trifolium repens) or in a multispecies mixture with additional legumes, bromegrass (Bromus willdenowii) and forbs. During each 9‐day experiment, botanical composition, milk production and faecal and urine composition were measured. Milk solid (MS) yield for the control ryegrass, high‐sugar ryegrass and tall fescue grass types averaged, respectively, 1.53, 1.64 and 1.70 kg MS cow−1 day−1 for a binary mixture sward, compared with 1.65, 1.54 and 1.53 kg MS cow−1 day−1 for a multispecies sward. Legume content influenced milk production more than the number of species present in a mixture. There was lower urine N concentration from a multispecies sward compared with a binary mixture. Urine N concentration of cows grazing the control ryegrass, high‐sugar ryegrass and tall fescue grass types averaged, respectively, 4.6, 5.3 and 6.8 g N L−1 for a binary mixture, compared with 4.1, 3.9 and 3.9 g N L−1 for a multispecies mixture. Feeding dairy cows on multispecies swards containing forbs presents an opportunity to reduce N losses without compromising milk yield.