Radlkofer (E.P. IV, 165: 1–1539 (1931–4)) devoted a lifetime to the monographing of the family. Recently Leenhouts, in preparation for Flora Malesiana, has revised many genera and reappraised the tribes erected by Radlkofer. In 1976 (Muller & Leenh. in The evolutionary significance of the exine) he suggested that Radlkofer’s more primitive tribes, the Sapindeae, Thouineae and Paullineae, are actually the most derived, and the Dodonaeae are not derived but a relict group. In effect this reverses the sequence of Radlkofer, and this is the order adopted here. Leenhouts (1976) further considers that the Aceraceae and Hippocastanaceae are no more than tribes of the Sapindaceae. [This is probably basically correct but these groups are of such horticultural importance I would not wish to tamper with their position – B.V.]The flowering biology of the Sapindaceae in East Africa is very poorly known and field observations are urgently needed for all genera and species. There are two main classes of sexual arrangement: dioecy, generally constant for a genus (but Allophylus has only some dioecious species) and monoecy. In the latter the tree has mass-flowering of all its inflorescences simultaneously, each inflorescence starting with male flowers which drop off by abscission of the pedicel shortly after anthesis. The next flowers to open are female, which if fertilised remain on the inflorescence, which then reverts to male flowers, sometimes in large numbers and continuing for a long season. However, this sequence may be imperfect through one or more stages being absent in any particular individual in any given season. The sex of the flower bud appears to be determined at least partly by environmental factors and at a late stage, and the sex of an immature inflorescence is impossible to determine in those genera where flowers are morphologically hermaphrodite – more than half those in East Africa. The term staminode is used here for a stamen which fails to dehisce. However, a small percentage of pollen in the staminodes of at least one species has been shown to be fertile.The structure of the fleshy seed-appendage has been the subject of some disagreements. Van der Pijl (Acta. Bot. Neerl. 6: 608 (1957)) argues that the structure is not homologous with the aril in other families and should be called an ‘arillode’. Corner (The seeds of Dicotyledons, 1976) disagrees and continues to use the term aril, while considering it sarcotestal in origin in some genera. The term aril will be used here regardless of ontogeny. Fruits of several species are quite unknown and should be looked for in the field.The East African genera belong to the following subfamilies and tribes:Subfamily Dodonaeoideae ( Dyssapindaceae of Radlkofer)Tribe Dodonaeeae – 1, Dodonaea Tribe Doratoxyleae – 2, Filicium; 3, Zanha Tribe Harpullieae – 4, Majidea Subfamily Sapindoideae ( Eusapindaceae of Radlk.)Tribe Cupanieae – 5, Aporrhiza; 6, Lychnodiscus; 7, Blighia; 8, Eriocoelum; 9, Haplocoelopsis Tribe Nephelieae – 10, Stadmania; Litchi; Nephelium; 11, Pappea Tribe Schleichereae – 12, Macphersonia; 13, Lecaniodiscus; 14, Haplocoelum; 15, Camptolepis Tribe Melicocceae – Tristiropsis; Melicoccus Tribe Lepisantheae – 16, Lepisanthes; 17, Glenniea; 18, Placodiscus; 19, Pancovia; 20, Chytranthus Tribe Sapindeae – 21, Deinbollia; 22, Sapindus Tribe Thouinieae – 23, Allophylus Tribe Paullinieae – 24, Cardiospermum; 25, Paullinia Several species have been or are cultivated in East Africa, including several well known fruits; all are included in the keys below. Cultivated species belonging to genera dealt within the text will be found there.