In attempting to arrange East African Papilionoïdeae so that, as far as possible, like genera may be near one another, the authors have not found either the classical arrangement by Bentham in “Genera Plantarum” (1865) or that by Hutchinson in “The Genera of Flowering Plants” (1964) wholly satisfactory. Bentham places Millettia and allied genera far from Lonchocarpus and its allies because the pod dehisces in the former but not in the latter group, though more than one good botanist has described a Millettia in Lonchocarpus or vice versa and in other parts of the family species with dehiscent and others with indehiscent fruits are included in the same genus. Both authors attach too great importance to transverse splitting of the fruit, a character combining separate dispersal of the seeds with the protection afforded by an indehiscent fruit which has, in our opinion, been evolved separately in at least three parts of the subfamily. This error led Bentham to place such plants as Antopetitia far from the closely related Loteae and Hutchinson to place Pseudarthria far from Desmodium, and Ormocarpum and its allies far from Dalbergia. We follow Hutchinson in thinking that the most primitive Papilionoïdeae are tropical trees or shrubs with pinnate leaves and free uniform stamens, in particular those which lack the usual differentiation of the corolla into standard, wings and keel. Like him, therefore, we start with Cadia and Dicraeopetalum followed by the remaining Sophoreae. Next are placed those tribes which are also tropical, woody and pinnate-leaved, then those which are tropical, often herbaceous and pinnate-leaved, then those which are tropical, mainly herbaceous and 3-foliolate. Tribe 11, the Genisteae, taken in a broad sense, is partly tropical, partly temperate, mainly herbaceous and 3-foliolate; it should almost certainly be subdivided, but to what degree is not yet clear. It is followed by two temperate, herbaceous, 3-foliolate tribes and lastly by 4 temperate, herbaceous, pinnate-leaved tribes. This arrangement is summarized in table 1.While admitting, as they must, that in principle physiological are as suitable as morphological characters for taxonomic use, many herbarium botanists will think us eccentric in attaching so much importance to the adaptation of groups to temperate as distinct from tropical climates. There is a feeling that nothing that cannot be seen in an unlocalized herbarium specimen should be used. Field workers, on the other hand, may be glad to find high-altitude genera grouped together. Our feeling that this climatic distinction is important has grown as we have studied the subfamily and has sometimes received unexpected confirmation. Column G of table 1, for example, shows the distribution of stipels in the Papilionoïdeae. It will be seen that they occur, though sometimes rarely, in all but one of the tropical tribes while they are absent in all the temperate tribes except the Psoraleeae where they are rare. Column H, taken from Dormer in New Phytol. 45: 146, table 1 (1946), shows the almost invariable presence of a foliar pulvinus, situated at the base of the petiole, in tropical tribes, but absent in temperate ones (a pulvinus is always present at the base of each leaflet, so 1-foliolate leaves are omitted from consideration). Dormer found other anatomical characters, such as open and closed vascular systems, rather closely correlated, but was unaware of the significance of his results in relation to the latitudinal centres of the groups he listed. In recent years much phytochemical data has been accumulated and its bearing will be more readily assessed when the review by Boulter, Harborne and Turner, “Comparative Phytochemistry and Chemotaxonomy of the Leguminosae”, is published shortly. Data from surveys of alkaloids and flavones seem to support the sort of evidence reported by Bate-Smith in Proc. Linn. Soc. 169: 203, table 2 (1958), for leuco-anthocyanins, which are generally present in tropical tribes and often absent or rare in temperate tribes (see table 1, column I). The basic chromosome numbers, taken from Turner & Fearing in Am. Journ. Bot. 46: 55, fig. 28 (1959) and the references they cite, are shown in column J. The numbers 10 and 11 are general in the tropical tribes, except Indigofereae and Sesbanieae, whereas 8, 7 and 6 predominate in the temperate tribes, except Psoraleeae. Theoretical justification for this arrangement may be drawn from the probable history of the subfamily. Although early Papilionoïdeae may have established themselves more than once in temperate regions much evolution from one or more of these pioneer stocks must have taken place there. It is more probable than not that any particular temperate group has been derived from another temperate group rather than directly from a tropical ancestor. A high degree of affinity between temperate groups is thus in accordance with expectation.