This article develops a framework for assessing thought experiments in normative political theory. It argues that we should distinguish between relevant and irrelevant hypotheticals according to a criterion of modality. Relevant hypotheticals, while far-fetched, construct imaginary cases that are possible for us, here and now. Irrelevant hypotheticals conjure up imaginary cases that are barely conceivable at all. To establish this claim, the article interrogates, via a discussion of Susan Sontag and Judith Butler’s accounts of representations of violence, the frames through which hypotheticals construct possible worlds, and concludes that some frames are better than others at sustaining a link with the world as we know it. Frames that disrupt this link can be charged with failing to offer action-guidance.