Two-sided ads often are more credible than one-sided ads because they admit that the advertised brands have shortcomings. Findings about the effectiveness of such ads have been inconclusive, however, perhaps because a critical moderator has been ignored. The author found that a two-sided ad was more effective than one-sided ads only when negatively correlated (vs. uncorrelated) attributes were featured. The brand's unfavorable positioning on the negatively correlated secondary attribute per se (i.e., "correlational inferences"), as well as the advertiser's honesty (i.e., "correspondent inferences"), jointly enhanced judgments of the brand on the primary attribute and thus overall brand evaluations.