Natural forests managed for timber production conserve tropical biodiversity, but how much? In a lowland wet forest of northeastern Costa Rica, we studied bees and understory flowers in regenerating forest plots logged 6-8 years previously, including silviculturally refined plots where non-timber species were removed. The experiment had a complete randomized block design with 3.24 ha plots (n = 3 per treatment) surrounded by 40 m buffer strips. Our four-month comparative study covered the major flowering period. Overall, 69 species monitored for visitors at 629 individuals demonstrated bees visited and likely pollinated 55 to 80% of understory plants[-]a mixture of primary forest species in shaded microsites (e.g., palms) and pioneers associated with relatively high insolation (e.g., certain Melastomataceae, Rubiaceae and Solanaceae). Small Halictinae predominated among the bees at flowers; 62% of the 21 species were undescribed. Few notable differences were found among plots of forest drastically altered by culling of non-timber species, and naturally regenerating forest. Such control and refined plots were similar in numbers of species flowering, richness of bees, visitor species per plant, and insolation. Both plot groups had similar complements of halictine species, as well as relatively low levels of foraging specialization by bees evidently nesting, as well as foraging, within plots. Bees were mainly active in shaded microhabitats of closed forest, but the small proportion of relatively insolated plants was heavily visited. Rarity of both native bees frequent in sunny, exposed sites (e.g., Megachile, Centris, Xylocopa) and of African honey bees, is evidence for existence of both a distinctive guild of understory bees and their perception of conditions in logged forests as similar to those of mature forest. Inglés