One of the most basic topics in social psychology is the way one agent influences the behaviour of another. This paper will focus on threats, which are an intensified form of attempted behavioural influence. Despite the centrality to the project of social psychology, little attention has been paid to threats. This paper will start to rectify this oversight. It reviews early examples of the way social psychology handles threats and highlights key limitations and presuppositions about the nature and role of threats. By contrast, we subject them to a programme of empirical research. Data comprise video records of a collection of family mealtimes that include preschool children. Threats are recurrent in this material. A preliminary conceptualization of features of candidate threats from this corpus will be used as an analytic start point. A series of examples are used to explicate basic features and dimensions that build the action of threatening. The basic structure of the threats uses a conditional logic: if the recipient continues problem action/does not initiate required action then negative consequences will be produced by the speaker. Further analysis clarifies how threats differ from warnings and admonishments. Sequential analysis suggests threats set up basic response options of compliance or defiance. However, recipients of threats can evade these options by, for example, reworking the unpleasant upshot specified in the threat, or producing barely minimal compliance. The implications for broader social psychological concerns are explored in a discussion of power, resistance, and asymmetry; the paper ends by reconsidering the way social influence can be studied in social psychology.