This paper is about face-threatening acts (FTAs). It upholds the usefulness of the general concept as first introduced by Brown and Levinson but presents a different view of (1) what an FTA is and how it arises and (2) how to gauge the relative severity of one. Regarding (1), it argues that no act is intrinsically face-threatening and that FTAs cannot be equated with speech acts. They can only be identified in the context of the ongoing interaction. An FTA, it is suggested, is simply any move which predicates a change in face. However, its ultimate identification rests with interactants. With regard to (2), the paper demonstrates that Brown and Levinson's ‘weightiness’ formula does not work. It is suggested that the severity of an FTA is the product of two factors: the amount of face-change it predicates and the amount of salience accorded to face at the time. But the values of these factors are also ultimately determined by participant reactions. Both matters are addressed largely through the close examination of one particular moment in one particular encounter.