ABSTRACT Information on the recovering of areas of gravel extraction of the Caatinga biome are scarce. Thus, the objective of this work was to assess the soil fauna as bioindicator of soil quality, evaluating areas with addition of topsoil (WAT) and control areas (CWT) without the addition of topsoil used for forest plantations, which were implemented for recovering gravel extraction areas. A native forest of the Caatinga biome (NFC) was taken as a reference. Ten traps with three replicates were installed in each area, which were evaluated in a randomized block design in split-plot arrangement at the end of the rainy season of different years, 0, 1, 3 and 6 years after the experiment implementation. A total of 45,740 specimens were captured. These specimens were from 36 taxonomic groups. The Acari, Diptera, Entomobryomorpha, Formicidae, Poduromorpha and Symphypleona were the predominant groups in all areas (WAT, CWT and NFC). Nine groups (25% of the total) had restricted occurrence. The fauna richness and diversity were higher in the NFC, followed by the WAT and CWT. The uniformity and total abundance of the soil fauna community showed no defined patterns. The abundance of most groups was inhibited in the treatments evaluated, compared with NFC, however, this adverse effect was more significant in CWT compared with WAT. The CWT and WAT had high similarity and both had very low similarity with NFC. The ecological complexity of soil fauna community was higher in NFC. The soil fauna was efficient as a bioindicator of soil quality, which was higher in WAT, compared with CWT.