We developed distributional models for 90 threatened bird species in China, and used heuristic complementarity algorithms to prioritize areas for conservation. The pixel-based area selection prioritized 20 areas for protection, which covered all species analysed. Area selecting for endangered species based on the existing biosphere reserve system included only 37-62 species in eleven reserves, leaving 28-53 species unprotected. We employed algorithms for area selection based on species richness and rarity and obtained two views of a most efficient reserve network. We used the distributional hypotheses to identify additions to the reserve system that would improve its effectiveness substantially. The pixel-based area selection approach includes species much more efficiently on a per-area basis, and thus offers exciting perspectives for improved protection of the country's endangered avifauna.