The revision was undertaken because the latest monograph by Pax & Hoffmann, dating from 1912, did not provide a satisfactory key, and because since that time a very large amount of new material has been collected and several new species were described. In the present revision 12 species have been recognized, among which 2 are new; 7 names have been reduced to synonymy. One species (E. eglandulosum Pax & Hoffm.) is here for the first time excluded from the genus and from the Euphorbiaceae and was found to be conspecific with Sterculia macrophylla Vent. (Sterculiaceae). The variability of the characters used by Pax & Hoffmann for infrageneric and for specific delimitation appears to be far greater than formerly understood. This was already stressed by Corner (1939) in his review of the Malayan species. For example, the peltation of leaves has appeared to be constant in some species, but not in others. The same can be said about myrmecophily — ants inhabiting hollow twigs — which is constant in 2 species, but is facultative in 2 other ones. This mild form of ‘parasitism’ is encountered only from Celebes eastwards as far as the Solomons (doubtful in Fiji). The variability in number of cells of the ovary appears also to be more complicated than formerly accepted on the strength of observations in a limited number of specimens. The occurrence of stellate hairs, and of indumentum in general, has not emerged as important for taxonomy. Remarks on this are made under E. diadenum. The number of basal nerves and the occurrence and position of the glands is also liable to variation and is constant in a few species only. Because of the variability of characters indicated above, the affinity of the species is reticulate and does not allow the distinction of clear infrageneric taxa. A difficulty encountered is that either flowers or fruits of several species are not or inadequately known and that some species are known from one specimen only. Though in 1867 Teysmann & Binnendijk mentioned the occurrence of polygamy in E. moluccanum, a ♀ tree and a tree with ♂ and some bisexual flowers, their observation was never again recorded in literature. I have not been able to confirm their observation in the abundant material of this species at hand. However, I have found that two species, E. banghamii and E. ronaldii from Sumatra and the Malay Peninsula respectively, possess bisexual flowers. Both are known from one collection only and more material must show the constancy of this feature which is almost unique in the Euphorbiaceae; recently it has also been recorded for two species of Aporosa by Airy Shaw. In collecting Endospermum one should always be aware of the necessity to preserve specimens with ♀ flowes and/or fruit, as sterile material or ♂ specimens are very difficult to identify. It is highly desirable that observations on variability and polygamy are made in the field or in botanical gardens.