Abstract Livestock farming in Great Britain (GB) faces multiple pressures. Yet, grassland managed for livestock is the most extensive habitat in GB and is key to cultural landscapes and their biodiversity and soil health. This study analysed a nationally representative dataset of over 940 large (200 m2) Neutral (agriculturally semi‐improved) and (agriculturally) Improved Grassland plots from the GB Countryside Survey (CS) to assess relationships between key grassland sward and soil variables. Analysis also looked at how these variables changed over time as plots switched between these grassland types. Data from grassland plots managed by Pasture‐Fed Livestock Association (PFLA) farmers were compared to CS plot data to assess the impacts of their practices on these variables. Plant species richness in Neutral grassland types in CS plots was positively associated with total soil invertebrate abundance (total taxa) and soil N and C and negatively associated with soil P. There were negative relationships between the covers of Lolium sp. (Neutral only), legumes and forbs and soil C and moisture variables. Grassland swards on PFLA member farms were characteristic of Neutral grassland. PFLA plots were more species rich and contained more legume and forb species and lower proportions of Lolium perenne than those on Improved Grassland. Vegetation height was greater in PFLA plots than in CS plots of either Improved or Neutral Grassland. Unlike CS Neutral Grassland plots, soil properties in PFLA plots were not significantly different from those for Improved Grassland for any measured variable (soil carbon concentration [C], bulk density, pH, nitrogen [N], phosphorus [P]). Higher species richness in grasslands is associated with positive measures of soil health. PFLA plant communities contain relatively high species richness and tall vegetation, which is positive for biodiversity. Lack of positive measures of soil health associated with higher species richness recorded in PFLA grassland (as opposed to CS grassland) may reflect time lags in soil responses to management, as evidenced through an analysis of the impacts of land‐use change over time on CS plot characteristics. Our findings indicate that pasture‐fed livestock approaches may be beneficial for grassland and wider ecosystems.