Abstract We respond to criticism of our earlier paper where we report Australia‐wide declines in fisheries catches that parallel the declining trends in fish populations observed underwater, and we highlight concerns about the low levels of precaution applied when regulating fisheries catches using the avoidance of recruitment failure approach. Most fished species worldwide lack the data needed for accurate stock status assessments, and consequently exploitation of these species should be managed with high precaution. For the relatively few species and stocks with individually modelled assessments, the errors associated with model output are extremely large as a result of the multiplicity of confounding factors (including effects of changing climate, technological advances that increase catch efficiency, fisher behaviour, interactions with other species, and changes in habitat quality), and the compounding of error introduced by subjective assumptions in multiple parameter estimates. The magnitude of this assessment uncertainty appears to be rarely recognized and incorporated into management decisions. Given the difficulties in accurately predicting and managing fishing impacts, including species interactions across space and time, a well‐designed set of no‐take marine reserves is critically needed. Although not a universal panacea, an effective global network of marine reserves arguably represents the most efficient and publicly acceptable next step – in addition to greenhouse gas reduction – towards solving the unfolding global dilemma confronting fish populations and ocean ecosystems.