Fences that exclude alien invasive species are used to reduce predation pressure on reintroduced threatened wildlife. Planning these continuously managed systems of reserves raises an important extension of the Single Large or Several Small (SLOSS) reserve planning framework: the added complexity of ongoing management. We investigate the long-term cost-efficiency of a single large or two small predator exclusion fences in the arid Australian context of reintroducing bilbies Macrotis lagotis, and we highlight the broader significance of our results with sensitivity analysis. A single fence more frequently results in a much larger net cost than two smaller fences. We find that the cost-efficiency of two fences is robust to strong demographic and environmental uncertainty, which can help managers to mitigate the risk of incurring high costs over the entire life of the project.