A single ecosystem can exhibit great biogeographic and environmental variability. While a given ecological driver might have a strong impact in one region, it does not necessarily hold that its importance will extend elsewhere. Coral reefs provide a striking example in that coral communities have low resilience in the Atlantic and remarkable resilience in parts of the species-rich Pacific. Recent experimental evidence from the Atlantic finds that fishing of large herbivorous fish can be a strong driver of coral resilience. Here, we repeat the Atlantic experiment in the highly resilient forereef of Moorea (French Polynesia), which has repeatedly recovered from disturbances. A combination of cages, fish deterrents (FDs), and controls allowed us to simulate the consequences of fishing large herbivores on algal assemblages, coral recruitment, and the demographic rates of coral juveniles. We find that the impacts of removing large herbivorous reef fish vary with early coral ontogeny. Reduced herbivore access led to a modest macroalgal bloom and reduction in coral recruitment. However, larger juvenile corals (> 1 cm diameter) survived better and grew faster under these conditions because of a reduction in corallivory. To determine the net impact of losing larger herbivorous fish, we combined experimental results with estimated demographic parameters in an individual-based model. Simulating coral recovery trajectories for five years, we find that protecting larger reef fish led to better recovery in 66–99% of simulations, depending on underlying assumptions (with the more credible assumptions being associated with greater likelihood of net positive impacts). While we find that fishing effects are detrimental to corals in both the Atlantic and Pacific systems studied, the nature of the interactions varied markedly. In the identical previously-published study in the Atlantic, macroalgae exhibited a rapid bloom and caused a sufficiently large reduction in coral recruitment to force a predicted ecosystem shift to an alternative attractor. The commensurate macroalgal bloom in Moorea was weak yet the corals were two orders of magnitude more sensitive to its presence. We do not suggest that a reduction in recruitment in Moorea will lead to alternative attractors but the long-term risks of a reduction in recovery rate are cause for concern as rates of coral mortality are projected to increase. The emerging picture is that Pacific reefs are less likely to experience macroalgal blooms but are surprisingly sensitive to such blooms if they occur.