We often react ambivalently to a piece of music, simultaneously experiencing both sadness and happiness, and attributing both emotions to the musical content. Two experiments were conducted (1) to empirically test for ambivalent emotions and (2) to investigate the amount of musical information necessary to elicit such emotions. In experiment 1, synthesized musical excerpts were manipulated in tempo (fast/slow) and mode (major/minor). By using unipolar scales, listeners could independently rate how (1) happy and (2) sad the music made them feel, as well as the (3) happiness and (4) sadness perceived in the music. Regarding perceived emotion, happiness was higher in major-fast excerpts, sadness was higher in minor-slow excerpts, and mixed emotions were reported (raised happiness and sadness ratings) for ambivalent music (major-slow and minor-fast). Ratings of felt emotions were similar, except that sad music (minor-slow) was experienced as ambivalent, happiness often nearly equalling the amount of sadness. Furthermore, the liking of sad excerpts positively correlated (r=.59) with the experienced happiness in the same excerpts. The results help to understand the enjoyment of sad music by suggesting that feelings of happiness are a desired emotional outcome. In experiment 2, the main results of experiment 1 were replicated with shorter musical excerpts (0.5-1.5s), suggesting that ambivalence is an immediate and effortless emotional response.