Forest restoration often involves monitoring programmes to determine whether biodiversity levels and ecosystem services have changed over time. This study investigated changes in ground-hunting spider communities (families Ctenidae, Lycosidae and Zodariidae) in an actively restored forest, an unrestored gravel mine, and two alternative land-use types (agroforestry system and an arable field) to assess whether a two-decade post-mine restoration programme has been successful in restoring biodiversity to levels of a reference natural forest. The overall activity density of ground-hunting spiders (based on both juveniles and adult specimens) was highest in the natural and the restored forest in the dry season and lowest in the arable field and agroforestry system in the wet season. The inverse Simpson index was highest at the gravel site in the wet season, followed by natural forest in both seasons and lower values in the restored forest. The community composition of spiders differed significantly between land-use types (open versus forest habitats) and the interaction between land use and season also differed significantly. The species Pardosa injucunda and Trochosa gentilis dominated the communities in the restored forest, but Africactenus monitor dominated the natural forest and Hogna gratiosa dominated communities in the gravel site. Surprisingly, active forest restoration promoted the activity density of ground-hunting spiders displaced by mining activities to levels even higher than in the reference natural forest after two decades. However, the community composition of the restored forest was more similar to the agroforestry system than to the natural forest. These results highlight the benefits of restoring former mining sites but also show the trade-offs in terms of restoration goals, as natural forest biodiversity of spiders was not achieved after 20 years.