Natural variation in longevity among populations of the same species, and between species and genera was investigated to inform seed-collection strategies. Seed longevity for 30 wild Australian populations was measured with a controlled ageing test. The populations were represented by eight species from three genera, namely Minuria (Asteraceae), Wahlenbergia (Campanulaceae) and Plantago (Plantaginaceae), each collected from up to eight different locations. Seed-survival curves were fitted by using the equation v = K_i + p/σ, which allowed comparison of the initial population viability (K_i), the population distribution of seed life spans (σ), and mean seed longevity (P₅₀, calculated as K_i × σ). At a genus level, the average P₅₀ indicated that M. integerrima (DC) Benth. is the longest-lived, Wahlenbergia is intermediate and Plantago is the shortest-lived. However, there was also variation in P₅₀ values among populations of most species. Some species had the same σ value for all populations, e.g. all eight populations of W. communis Carolin had the same σ value, with the differences in K_i causing the variation in P₅₀. This consistency in σ existed even though seedlots were collected from diverse locations, with mean annual rainfall ranging from 180 to 840 mm. In comparison, for the six seedlots of W. gracilis (G.Forst.) A.DC., a large difference in σ as well as K_i led to the variability in P₅₀, with some indication of a possible correlation between annual rainfall and P₅₀ or σ in some species. A relationship between variation in σ and the breeding system is proposed for Wahlenbergia. The data show that it can be risky to expect accurate prediction of seed longevity for a wild species on the basis of survival data from a single collection.