For any proposition p, it may sometimes occur that a person is not quite accurately describable as believing that p, nor quite accurately describable as failing to believe that p. I describe such a person as in an ‘in-between state of belief’. I argue for the prevalence of in-between states of believing, and assert the need for an account of belief that allows us intelligibly to talk about in-between believing. I suggest that Bayesian and representationalist approaches are inadequate to the task, and that a Rylean dispositional account of belief might do the trick.