In the present study, a two-dimensional model of human emotional reactions is used to predict auditory perception. The models’ discriminant validity is tested in an experiment where one group of participants use a two-dimensional emotion measure to rate their reactions to sounds systematically varied in tone and noise spectral level content. Another group of participants rated the same sounds using a standard one-dimensional annoyance measure. The results showed that both the one- and two-dimensional measures discriminated between reactions to different sounds. Regression analyses showed that the two-dimensional measure tapped aspects of human experience not covered by the annoyance measure. In addition, it was shown that only modifications of the fundamental frequency and the overall noise spectral level, but not modifications of harmonics, had a marked effect on emotional reactions. The implications of these findings for both auditory emotion research and noise control engineering are discussed.