Marine Protected Areas are usually static, permanently closed areas. There are, however, both social and ecological reasons to adopt dynamic closures, where reserves move through time. Using a general theoretical framework, we investigate whether dynamic closures can improve the mean biomass of herbivorous fishes on reef systems, thereby enhancing resilience to undesirable phase-shifts. At current levels of reservation (10-30%), moving protection between all reefs in a system is unlikely to improve herbivore biomass, but can lead to a more even distribution of biomass. However, if protected areas are rotated among an appropriate subset of the entire reef system (e.g. rotating 10 protected areas between only 20 reefs in a 100 reef system), dynamic closures always lead to increased mean herbivore biomass. The management strategy that will achieve the highest mean herbivore biomass depends on both the trajectories and rates of population recovery and decline. Given the current large-scale threats to coral reefs, the ability of dynamic marine protected areas to achieve conservation goals deserves more attention.