Herbivory is one of the key processes shaping life history traits linked to plant fitness and the leaf-cutting ants are amongst the most voracious and polyphagous herbivores of the Neotropics. They extensively harvest aboveground plant parts to grow their symbiotic fungus, including vegetative and reproductive tissues. Although leaf-cutting ant herbivory is expected to reduce plant reproductive success, the potentially different outputs of damage to vegetative and reproductive parts are still poorly known. This study investigates the influence of Atta opaciceps herbivory on the reproductive success of three plant species native to the Caatinga dry forest: Croton argyrophylloides (Euphorbiaceae), Cenostigma pyramidale, and Indigofera suffruticosa (Leguminosae). We selected 10 adult and active nests of A. opaciceps based on the presence of individuals of the three focal plant species in their vicinity. For each colony, we marked four individuals of each plant species (totaling 12 individuals per colony), which were then subjected to one of four herbivory treatments: (1) ant herbivory on both reproductive and vegetative parts, (2) manually simulated herbivory on the reproductive parts alone, (3) ant herbivory exclusively on vegetative parts, and (4) no herbivory. From January to April 2020, we monitored all 120 individuals, recording the number of inflorescences, flower buds, flowers, fruits, and the proportion of flowers maturing into fruits (fruit set, as a proxy of reproductive success). Results revealed variations in reproductive structure production and divergent responses to leaf-cutting ant herbivory among our focal species. In terms of inflorescence, flower bud production, and fruit set C. argyrophylloides showed reduced numbers when subjected to higher levels of total herbivory (reproductive and vegetative parts), but increased inflorescence and flower bud production when plants suffered only leaf damage. Cenostigma pyramidale individuals under partial herbivory displayed lower flower bud numbers than protected ones. In contrast, I. suffruticosa displayed conflicting results, with fully exposed plants producing more inflorescences than partially exposed counterparts and more flowers than leaf-damaged and protected individuals. Despite the varied effects on different plant parts across all three study species, the overall reproductive success (i.e. fruit set) of only one species (C. argyrophylloides) was significantly reduced by leaf-cutting ant herbivory. The resilience of the two other plant species to leaf-cutting ant pressure may be attributed to compensatory mechanisms, shedding light on the intricate interplay between herbivores and plants in the Caatinga dry forest. As leaf-cutting ants proliferate in response to anthropogenic disturbances, the patterns described in this study may become more pronounced, potentially affecting the organization of plant communities in the Caatinga dry forest.