The Pacific Islands staples (mostly root and tuber crops) have been clonally distributed over wide geographical distances. Over the last three decades, molecular markers studies have confirmed that the bottlenecks induced by this process resulted in narrow genetic bases exposing smallholders to increased vulnerability. Farmers can select new cultivars when capturing new morphotypes clonally from seedlings or sports. When interesting, clones are easily exchanged and distributed. However, because of the present climatic, environmental and socio-economic changes, improved germplasm is now urgently needed. These species share common biological traits: they have variable ploidy levels but are predominantly allogamous and highly heterozygous. Breeding is a slow process. When available, the distribution of selected clones is constrained by the large number of smallholders, their geographical isolation, the absence of a 'seed' industry, poor national dissemination and distribution systems, their low multiplication rate and strict international regulations. The question is: how can we develop an efficient system to distribute new cultivars, considering climatic uncertainties, the geographical constraints and the characteristics of the species involved? This paper presents a review of the Pacific traditional crops genetic diversity, breeding progresses and the difficulties of the existing germplasm distribution and utilization system. A few practical solutions to strengthen farmers' capacity to adapt to forthcoming changes are proposed. (Résumé d'auteur)