This paper evaluates Kieran Healy's recent contribution to the structure–agency debate. Supervenience, I argue, has multiple uses, it entails different ontological perspectives depending on which entities it is applied to and which conditions are placed upon subvening and supervening entities. Healy's use of supervenience is unclear. On the one hand, applied to individual–society relations it does nothing more than restate the trivial truth: no people – no society. On the other hand, if supervenience is to be applied to structure–agent relations the consequence is extreme voluntarism. In either case it simply fails to address Healy's key concern: conceptualising social constraint. I then argue that an alternative way of grasping structural constraint in the present might be to view past-tense ‘activity dependence’ as ‘Cambridge events’.