In a seminal contribution, Fisher argued how distastefulness could incrementally evolve in a prey species that was distributed in family groups. Many defended prey species occur in aggregations, but did aggregation facilitate the evolution of defence as Fisher proposed or did the possession of a defence allow individuals to enjoy the benefits of group living? Contemporary theory suggests that it can work both ways: pre-existing defences can make the evolution of gregariousness easier, but gregariousness can also aid the evolution of defence and warning signals. Unfortunately, the key phylogenetic analyses to elucidate the ordering of events have been hampered by the relative rarity of gregarious species, which in itself indicates that aggregation is not a pre-requisite for defence. Like the underlying theory, experimental studies have not given a definitive answer to the relative timing of the evolution of defence and aggregation, except to demonstrate that both orderings are possible. Conspicuous signals are unlikely to have evolved in the absence of a defence and aggregated undefended prey are likely to be vulnerable to predation in the absence of satiation effects. It therefore seems most likely that defence generally preceded the evolution of both aggregation and signalling, but alternative routes may well be possible.