In this article, we propose to deepen the extreme as a notion in order to characterize it independently of the ontologies underlying each research interested in extreme activity, event, environment, context, situation, organization... The objective is to suggest an ordered partition of important notions of the field. Thus, we propose a) to define the extreme as the association of three components: rupture, unknown and risk, b) to consider emergency, crisis and catastrophe as ingredients increasing the extreme and c) finally to make a partition between the constrained extreme and the deliberate extreme. In conclusion, we discuss the legitimacy of this initial framework: could we characterize the extreme(s) without a reference to the ontologies underlying the various researches?