In three experiments, we tested whether music listening evokes mixed emotional responding. Undergraduates listened to music that varied in tempo (fast or slow) and mode (major or minor), such that it had consistent happy (fast/major), consistent sad (slow/minor), or inconsistent (fast/minor or slow/major) emotional cues. Participants were asked to rate their reaction to the music on several pairs of semantic opposites (happy-sad, pleasant-unpleasant, like-dislike). Across experiments, fast tempi and major mode were associated with higher ratings on positive scales; slow tempi and minor mode were associated with higher ratings on negative scales. As predicted, "mixed" positive and negative responding was evident, but it was greater when tempo and mode cues conflicted only for mixed happiness and sadness. These response patterns were evident whether we asked listeners how the music made them feel or how it sounded, and whether the stimuli varied in a highly controlled or an ecologically valid manner.