Bas van Fraassen argues that inference to the best explanation (IBE) is a probabilistically incoherent rule (1989). Anyone following IBE is open to being Dutch booked. According to one of the most interesting and popular responses to van Fraassen’s argument, van Fraassen misrepresents IBE in probabilistic terms. With the proper probabilistic representation, it is claimed, IBE is not inconsistent with Bayesian rationality constraints. Building on the work of IBE’s proponents, I first propose a minimal account of what makes one explanation better than another. I then argue that, even on this minimal account, the alternative probabilistic model of IBE does not work, and hence fails to successfully respond to van Fraassen’s argument.