© 2015 The Linnean Society of London. Since the last classification of Orchidaceae in 2003, there has been major progress in the determination of relationships, and we present here a revised classification including a list of all 736 currently recognized genera. A number of generic changes have occurred in Orchideae (Orchidoideae), but the majority of changes have occurred in Epidendroideae. In the latter, almost all of the problematic placements recognized in the previous classification 11 years ago have now been resolved. In Epidendroideae, we have recognized three new tribes (relative to the last classification): Thaieae (monogeneric) for Thaia, which was previously considered to be the only taxon incertae sedis; Xerorchideae (monogeneric) for Xerorchis; and Wullschlaegelieae for achlorophyllous Wullschlaegelia, which had tentatively been placed in Calypsoeae. Another genus, Devogelia, takes the place of Thaia as incertae sedis in Epidendroideae. Gastrodieae are clearly placed among the tribes in the neottioid grade, with Neottieae sister to the remainder of Epidendroideae. Arethuseae are sister to the rest of the higher Epidendroideae, which is unsurprising given their mostly soft pollinia. Tribal relationships within Epidendroideae have been much clarified by analyses of multiple plastid DNA regions and the low-copy nuclear gene Xdh. Four major clades within the remainder of Epidendroideae are recognized: Vandeae/Podochileae/Collabieae, Cymbidieae, Malaxideae and Epidendreae, the last now including Calypsoinae (previously recognized as a tribe on its own) and Agrostophyllinae s.s. Agrostophyllinae and Collabiinae were unplaced subtribes in the 2003 classification. The former are now split between two subtribes, Agrostophyllinae s.s. and Adrorhizinae, the first now included in Epidendreae and the second in Vandeae. Collabiinae, also probably related to Vandeae, are now elevated to a tribe along with Podochileae. Malaxis and relatives are placed in Malaxidinae and included with Dendrobiinae in Malaxideae. The increased resolution and content of larger clades, recognized here as tribes, do not support the 'phylads' in Epidendroideae proposed 22 years ago by Dressler.