Male euglossine bees were sampled with chemical baits every two months from September 1997 to July 1999 at nine sites in the Desengano mountain range, Rio de Janeiro State, Brazil. Four sites were located in Atlantic Forest mature second growth, two sites in disturbed forest, and three sites in forest fragments of 200, 156, and 14 ha, respectively. We collected 3653 male euglossine bees from at least 21 species. Analyses of variance indicated no differences among the three habitat types for total number of bees, and 5 of the 6 dominant species. Bootstrapping indicated significant variation in species richness and diversity for some sites, but there was no clear indication of differences among habitats. Similarity as measured with the Morisita–Horn index was inversely related to distance between sites, but relatively high for most site combinations. These results suggest that the euglossine bee community in the three habitats was essentially the same. Although some species were associated with each habitat type, most appeared to respond to temporal local conditions. Our results do not support the hypothesis that forest fragmentation or habitat alteration reduces abundance and diversity of euglossine bees.