Population viability analyses (PVA) are central tools for the management of threatened populations. However, the parameterisation of effective PVA models is very demanding in high quality data, which are often impossible to collect on endangered populations. Here we propose the use of a generalisation strategy to bypass this limitation: management measures for an endangered metapopulation of the cranberry fritillary butterfly in the Netherlands are evaluated with RAMAS/GIS by using parameters estimated from a healthier metapopulation in Belgium. The Belgian metapopulation seems viable, with stable abundance and number of local populations, despite their erratic dynamics, whereas the Dutch metapopulation shows a continuous decline in the course of time, with many vacant habitat patches. Simulations of various scenarios indicated that (1) large scale restoration of habitat patches would be necessary to ensure long-term survival of the species in the Netherlands as not enough suitable habitats are currently remaining; and that (2) global warming is expected to put a major threat on both metapopulations by reducing the growth rate of this glacial relict species, and/or increasing environmental stochasticity (amplified climatic variations).