Global initiatives to recover degraded areas have intensified in recent years. However, the effectiveness of different reforestation practices in restoring habitat stability and essential resources for species over time is poorly known. We tested the degree of temporal stability of bee and wasp communities reestablishiment in different types of forest regeneration compared to intact primary forest in southern Amazonia. Using trap-nest sampling, we collected, during 12 months, information about the nesting species in six different types of treatments: pastures, teak reforestation, fig and mixed native species, natural regeneration forest, and forest primary. We observed that in all reforested treatments, the abundance and richness of both bees and wasps were greater than in pasture sites. However, not all reforestation strategies had the same degree of effectiveness in maintaining the stability of bee and wasp communities over time than observed in primary forest communities. The high temporal variation in teak structure promoted high instability in bee and wasp communities over time. In contrast, reforestation using perennial species, such as fig or mixed species, resulted in a greater temporal stability of bee and wasp communities, although dissimilarity to primary forest remained high. Communities in natural regeneration were the most similar to primary forest, with high similarity of species composition and high temporal stability across seasons. Our findings suggest that reforestation strategies that promote increased complexity of vegetation structure, and better resemble "natural" forests in the provision of habitat resources throughout the year, will better restore and maintain communities through time.