Sampling a whole flora or any significant subset for ex situ conservation purposes is a complex, long-term proposition. Thus, it is important to consider what constitutes an adequate sample not only for all taxa as a whole but also for each taxon individually as well as how to strategically schedule collection over time. There are five basic sampling questions: from which species to collect, from how many and which populations, from how many and which individuals, how many and what kind of propagules to collect, and, finally, at what point is the desired sample size too great for a population to bear in one year? There is no single correct sampling strategy or protocol. Each situation must be evaluated in the context of the particular purposes, goals, and uses for which the collection is being made as well as the nature of the sampling universe, the manner in which the samples will be stored and maintained, the period of time they will be stored, and whether the wild-collected samples are to be used directly or whether their numbers will be increased by agricultural growouts. Purposes include providing a long-term “insurance policy” against extinction in the wild and, in the near to medium term, supplying native plant material for small- to large-scale restoration purposes. The term “sampling universe” refers to the target taxa of interest: it could be the entire seed plant flora of Taiwan or particular subsets, such as endemics, higher-elevation plants, ecologically important taxa, rare plants, and so on. Plants with orthodox seeds are much easier to work with and are more economically stored as dried, frozen seeds in a seed bank than either plants with recalcitrant seeds or taxa maintained as growing plants.