A field mixtures experiment was conducted in Zürich over three years (2012-14) in which the sown legume proportion was 86, 50 or 14% in the seed mixtures and 100 or 0% in monocultures. In the spring of 2015, these swards were removed using herbicide and all plots were re-seeded with Lolium multiflorum L. monocultures, with the aim of studying soil transferred legacy effects of legumes on the performance of a following grass crop. The L. multiflorum L. was then harvested three times during 2015 (with the yield from each cut being summed for analyses) and once in spring 2016. Nitrogen (N) concentration from the combined harvests in 2015 was also recorded. A highly significant increase in the L. multiflorum L. annual biomass yield was detected in all treatments where legumes were the preceding crop compared to the non-legume monoculture in 2015 (P < 0.001). An effect continued into the following year, where treatments containing > 50% of sown legume proportion had a significantly greater biomass than the non-legume treatments (P < 0.01). Lastly, N concentration in the L. multiflorum L. sward from the 2015 harvest was also significantly increased in treatments with sown legumes over the non-legume monoculture (P < 0.05). These results explain how legume-induced effects on biomass could be sustained in mixed swards, even as the legume proportion declines to low levels, as had previously been observed in the COST 852 agro-diversity experiment.