Five mycoheterotrophic species of Cephalanthera have been proposed from tropical Asia. Until recently, all of them were only known from the type specimens, for which reason it has been difficult to judge if some of them were actually conspecific. The recent discovery of two large populations of C. exigua in Thailand made it possible to describe the range of variation of this species in much more detail, and we then found it to be well distinguished from the others. A phylogenetic analysis based on nuclear rDNA ITS sequences corroborated the systematic position of C. exigua at the generic level. Examination of the previously neglected type of C. pusilla from Myanmar lead to the conclusion that this taxon of disputed generic affinity likewise belongs to Cephalanthera, and that it is a clearly distinct species. Finally, we accept that also C. ericiflora from Laos as well as C. gracilis and C. calcarata from Yunnan are probably distinct species. Cephalanthera exigua grows in hill evergreen forest, but we found little floristic similarity between its two localities in Thailand (ISS=22.5%). Ectomycorrhizal trees of the families Fagaceae and Ulmaceae are probably the ultimate carbon source for C. exigua at both sites, as the mycorrhizal partners of this orchid (basidiomycetes of the family Thelephoraceae) obtain their carbon nutrition through ectomycorrhizal association with forest trees. Pollination experiments demonstrated that the nectarless flowers of C. exigua are not spontaneously autogamous. Based on the pollination syndrome and previously published observations on other species of the genus, we suggest that C. exigua has an insect-operated breeding system acting by deceit, and that the pollinators are probably solitary bees.