The evolution of dominance by the selection of modifiers of the phenotypes of deleterious mutations was proposed as a hypothesis by R. A. Fisher in 1928. It has been strongly criticized ever since, is regarded by many as having been made irrelevant by metabolic control theory, and most recently has been claimed to have been ‘falsified’ by H. A. Orr. Is it indeed not only obsolete but wrong? Its history is reviewed and its present status evaluated. We conclude (1) that it has a role as the explanation of the dominance found in many cases of selection through visual predation and (2) that the selection mechanism long claimed to be ineffective (the increase in frequency of a single modifier) will be effective under certain special conditions that may be different from those Fisher proposed.