What is curiosity? An attractive option is that it is a desire to know. This analysis has been recently challenged by what I call interrogativism, the view that inquiring attitudes such as curiosity have questions rather than propositions as contents. In this paper, I defend the desire-to-know view, and make three contributions to the debate. First, I refine the view in a way that avoids the problems of its simplest version. Second, I present a new argument for the desire-to-know view that focuses on ascriptions of the form ‘S is curious to ϕ’, which, despite their prevalence, have been ignored in the literature. Third, I examine the central motivation for interrogativism – the argument from metacognition, according to which animals can be curious yet do not have the metacognitive capacities required by desires to know – and argue that it rests on questionable assumptions about desires and attitude ascriptions.