A recent proposal by Quill Kukla and Mark Lance holds that surface appearances notwithstanding, gender ascriptions are closer to normative performatives than descriptions. As speech acts, they share more in common with pronouncing a marriage than a neutral description of a person, albeit this is not commonly recognized. This paper argues that the proposal faces a consistency problem. In order to affect social reality qua their illocutionary force, gender ascriptions must on average succeed. However, according to the authors most actual (...) second- and third person gender ascriptions in fact violate the success conditions of the speech act type, which include respect for the individual’s first-person gender ascriptions. Their main claim thus becomes that gender roles can be instituted as a series of misfires, though it is left open how this is possible. (shrink)