This research examines, across 2 studies, the interplay between the valence and arousal components of affective states and the affective tone of a target ad. In the first study, music was used to induce a pleasant or unpleasant mood, while controlling for arousal. Participants were subsequently exposed to an ad that either had a positive-affective tone or was ambiguous in its affective tone. As predicted, the valence of the affective state colored the evaluation of the ad in a mood-congruent direction, but this coloring effect occurred only when the ad had an ambiguous-affective tone. In the second study, the target ad had a clear positive or negative affective tone, and the valence and arousal dimensions of the mood state were manipulated independently. As predicted, the arousal dimension, but not the valence dimension, influenced ad evaluation. Ad evaluations were more polarized in the direction of the ad's affective tone under high arousal than under low arousal. This effect was more pronounced for self-referent evaluations (e.g., "I like the ad") than for object-referent evaluations (e.g., "The ad is good"), favoring an attributional explanation - the excitation transfer hypothesis - over an attention-narrowing explanation - the dynamic complexity hypothesis. Taken together, the results of the 2 studies stress the important contingency of the affective tone of the ad, when examining the effects of the valence and arousal dimensions of a person's affective state on ad evaluation. The results also provide additional insights into how and when affect serves as information in judgment processes.