Abstract Possible Worlds Theory has commonly been invoked to describe fictional worlds and their relationship to the actual world. As an approach to genre, the relationship between fictional worlds and the actual world is also constitutive of specific text types. By drawing on the notion of accessibility relations, different genres can be classified based on the distance between their fictional worlds and the actual world. Maître, Doreen. 1983. Literature and possible worlds. Middlesex: Middlesex University Press for example, in what is considered the first attempt to adapt accessibility relations from logic to literary studies, distinguishes between four text types depending on the extent to which their fictional worlds can be seen as possible, probable, or impossible in the actual world. Developing Maître’s work, Ryan, Marie-Laure. 1991a. Possible worlds and accessibility relations: A semantic typology of fiction. Poetics Today 12. 553–576, c.f. Ryan, Marie-Laure. 1991b. Possible worlds, artificial intelligence, and narrative theory. Bloomington: Indiana University Press) creates a comprehensive taxonomy of accessibility relations that may be perceived between fictional worlds and the actual world. This includes assuming compatibility with the actual world in terms of physical laws, general truths, people, places, and entities. Using her taxonomy, she then offers a typology of 13 genres to show how fictional worlds created by different genres differ from each other. As it stands, Ryan’s typology does not contain the genre of counterfactual historical fiction, but similar genres such as science fiction and historical confabulation are included. In this article, specific examples from counterfactual historical fiction are analysed to show why it is problematic to place these texts within the genres of historical confabulation or science fiction. Furthermore, as I show, Ryan’s typological model also does not account for some of the characteristic features of the genre of counterfactual historical fiction and as such the model cannot account for all texts within the genre. To resolve this issue, I offer modifications to Ryan’s model so it may be used more effectively to define and distinguish the genre of counterfactual historical fiction.