The spatial prioritization of biodiversity restoration sites on a national scale lags behind that of protected areas. We present a simple framework for spatial restoration prioritization using complementarity analysis based on the differences between past and present species distributions and restoration feasibility. In the framework, sites that were not only diverse or unique in the past (high restoration potential), but also have high restoration feasibility (e.g. low anthropogenic pressures) are ranked higher. We applied the framework to endangered bird species that breed in Japan for which distribution data were available from nationwide distribution censuses conducted in 1978 and 1998. The data were compiled in a 20-km grid resolution across Japan. We incorporated the degree of urbanization and number of farmers as a cost to optimize the sets of cells with regard to restoration feasibility. We successfully identified candidate sets of restoration cells, in which at most six can be allocated for each species under the constraint that the total area selected does not exceed 15% of the degraded cells. At most, 11 cells were allocated for each species under the constraint that restoration cells were within 15% of all ecosystem types. Comparison with the distribution of existing nature restoration sites revealed that only 9.3 or 9.7% (depended on the constraint) of high-priority cells overlapped cells including existing restoration sites. Our framework provides an objective, efficient method to obtain a testable plan for biodiversity restoration, which might contribute to the attainment of the 2020 target set by the Convention on Biological Diversity.