Persuasive messages can be interpreted in positive or negative ways, making it difficult to predict message effectiveness. We analyze theoretically and experimentally the role of causal reasoning in the interpretative process, as applied to striking real-world examples of persuasion. In our model, attributes and the relationships between them are mentally organized in a Bayesian belief network. Due to this structured mental representation, seemingly negative information can be viewed in a positive light, or vice versa. We conduct an experiment based on the car rental agency Avis and their famous slogan, “We're No. 2—that means we try harder,” and show that their ad's effectiveness can be predictably modulated by altering background beliefs. Thus, Bayesian reasoning can help explain counterintuitive impacts of persuasive messages.