In the wild and in cultivation, sago palm (Metroxylon sagu Rottb.) is mostly propagated vegetatively through its suckers. Generative propagation occurs to a limited extent; there are strong indications of sago palm being largely an obligatory cross-pollinator. This opens the possibility of the existence of an endless amount of clonal varieties of sago palm, which is confirmed by the endless amount of different names for locally distinguished varieties in the areas where sago palm occurs naturally. Sago palm's great variability is now threatened by the decrease in area under natural and semi-natural sago forest in the centres of diversity (New Guinea, the Moluccas) and by the sole attention to high-yielding, quickly-maturing varieties for plantation purposes. A comprehensive plan is required for the conservation of the existing germplasm for future selection and breeding. The problems and uncertainties with in-vitro and standing in-situ germplasm collections are such that gathering varieties from the wild and planting them in special regional germplasm gardens seems the only viable option. Issues of collecting and describing material are dealt with, including the evaluation of information from local informants. Centres of sago palm diversity seem to coincide with centres of language diversity. It is demonstrated, however, that differences in name cannot be attributed singly to differences in language. An extensive list is given of examples of locally distinguished varieties reported in literature, and of varieties encountered during the author's 1988–1990 research on sago in the Moluccas and during a two-month all-Indonesia study tour in 1992.