Gambling can be defined as a behavior where one wagers something of value to gain a large reward in a situation of a degree of uncertainty. Understanding gambling is critical as gambling disorder is a globally prevalent common mental health concern and an important public health issue. Given that gambling is often approached through cognitive models, it would be feasible to consider gambling disorder nearly as a token example of predictive processing. Curiously, however, relatively little has been written about problematic gambling from the viewpoint of predictive processing. One notable exception to this is Schwartenbeck et al. (2015) who argue that the cause of suboptimal behavior is the failure in the generative model of the world and exemplify it by modeling gambling behavior in relation to impulse control issues. In our chapter, we take their general position as our starting point and regard the cause of suboptimal behavior to be the gambler’s prior probabilities in the generative model of the world. From this, we illustrate how known real-life facts about the recreational and problem gamblers suggest that they have various prior beliefs, and this makes their gambling behavior rational from their own perspective. We end by considering the prospects that this new perspective on the rationality of gambling activities affords with regard to more informed treatment development and personalized treatment decisions.