In his treatment of the genus, Capuron (1969) followed the work of Leenhouts (1967) for other parts of the world, in which extremely broad species circumscriptions were proposed, broken up into numerous "races" (a category that is not accepted by the code of nomenclature, and has no nomenclatural standing). Capuron regarded all material from Madagascar and the Comoros to belong to a single species - the Asian Allophylus cobbe (L.) Raeusch. Within this enormously variable "species", Capuron recognized 20 different races. Today, this approach has largely been discredited, and it is doubtful that true Allophylus cobbe occurs in Madagascar. Nevertheless we judge that the entities described by Capuron mostly merit recognition at species level, and indeed 13 of them already have valid species names. We therefore threat the races as taxa at species level, pending a taxonomic revision of the genus for Madagascar. Certain other ms. names exist that we regard as probably representing distinct taxa at species level, these too we treat as species in this account, and in doing so we hope to provide a better understanding of the diversity of the genus in Madagascar and to encourage and facilitate the first steps towards the much needed revision.