The author supports the view that natural mortality always predominates over fishing mortality in fish populations not strictly linked to a specific substratum, fishing mortality thus being a subsidiary factor for changes in abundance. That statement is supported by long historical catch series available, deposition rates of fish scales on anoxic sediments and research on recruitment variability. Man cannot avoid changes in fish abundance in the long-term, because fishing is not the major factor controlling fish abundance in most cases. This fact should lead to changes in the meaning of the term “overfishing”. It could be defined as the very low level of biomass and CPUE for that phase of the lifespan when fishing mortality controls abundance, so that the curve of average biomass per recruit and CPUE versus fishing mortality levels off. moderate changes of effort in an overfishing situation would lead to undetectable changes of biomass per recruit and CPUE, much lower than those caused by recruitment variability. Management should then be made “on blind conditions”. Reductions in effort on an overfished situation should not lead to losses in yield. The author believes that an effort to better understand fish population biology should be of higher priority than the continuous attempts made to optimize the use of current assessment models.