Harvesting reduces fish populations. Using empirical data and stock-recruitment models, a biomass limit reference point may be set to maintain recruitment. This minimum is often <20% of the biomass that would be present without fishing. However, such a low biomass might be inadequate to sustain top predators, especially where an exploited fish stock is a key food for wildlife. Another limit reference point might be devised to account for food needs of predators. Empirical evidence shows that bioenergetics estimates of the quantities of food required by top predators are inadequate to set reference points. Top predators such as marine mammals and seabirds show population declines or breeding failures when food stocks remain far above the minimum amount these predators need to eat. The density of the prey field for foraging predators is probably a crucial factor affecting their foraging performance. Not all species of seabirds or marine mammals are equally vulnerable to impacts on their populations through food shortage. It is possible to identify the aspects of predator ecology most likely to require high densities of food to permit economically profitable foraging. It may be useful to select for study ‘sensitive species’ of seabirds and marine mammals; this might permit the development – from empirical studies – of reference points based on the ecology of these species. Such reference points for ‘sensitive species’ could be used in a precautionary way as proxies to protect the broad community of dependent wildlife.