Environmental indicators can be used to evaluate natural recovery processes. Although several studies have used ant assemblage metrics as environmental indicators, few try to explain the response of these insect assemblages to forest recovery. Thus, we carried out a vertical ant sampling design to verify the sensitivity of ant assemblages from different microhabitats (epigeic, arboreal and hypogeic) to natural forest recovery. We also verified whether changes to the diversity measures (species richness, composition and evenness) of these assemblages could be predictable and explained by changes in the resource quantity and conditions, and structural and resource heterogeneity that also occur along the natural forest recovery gradient. Additionally, we verified the potential of ant species as indicators of forest recovery. We sampled ants in a pasture and seven forest remnants with different recovery times that represent a forest recovery gradient along a chronosequence. Although, the non response of species richness and evenness, species composition changed along the recovery gradient and the dissimilarity between the ant species composition of pasture and forest habitats increased with the age of the forest remnant. This was supported by ant species indicators of pasture and forest areas pointed out by the IndVal test. The variation of ant species composition of the two groups of habitat types (pasture and forest) could be explained by the different responses of open habitat and forest specialist ants to the increase in vegetation cover along the recovery gradient. Therefore, species composition and indicator species by habitat type seems to be the most predictable data for the use of ants in environmental monitoring programs.