Setting aside networks of protected areas for conservation is urgently needed to counteract the current extinction crisis. Complementarity-based reserve selection algorithms have been developed in recognition that such a task needs to make the best possible use of the scarce resources available to conservation, maximising the return in terms of biodiversity protection. This project aims to contribute to the improvement of these algorithms, particularly using optimisation methods, to make them more applicable to practical reserve selection. In pursuing this objective, a number of different approaches are adopted. Using different exemplar data sets, I (i) explore methods for the evaluation of existing networks of protected areas; (H) develop guidelines for the selection of networks which are more robust to species temporal turnover, and present evidence that minimum complementary sets tend to select areas of ecological transition; (Hi) demonstrate how optimisation tools can be applied to maximise phylogenetic diversity, and present evidence that complementary sets maximising for taxonomic richness are adequate surrogates in representing phylogenetic diversity; (iv) demonstrate how species rarity influences complementary reserve selection across geopolitical boundaries; (v) provide guidelines for the application of reserve selection algorithms in areas with poor biological data; and (vi) investigate what should be adequate conservation targets for reserve networks representing plant and vertebrate species, in the tropical rain forests and at a global scale. I then put the results obtained in this thesis and other published literature in a broader context, analysing the explanations as to why reserve selection algorithms are failing to have an impact in conservation practice. This study demonstrates the flexibility of reserve selection algorithms as tools for the selection of complementary reserve networks, and proposes developments needed to improve their effectiveness as practical conservation planning tools.