Two experiments are reported which test the hypothesis that the disconfirmation of an expectancy is unpleasant. In Experiment I, Ss in two groups--Expectancy Confirmed and Expectancy Disconfirmed--tasted quinine and sugar solutions. Contrary to predictions, quinine was judged more pleasant when tasted unexpectedly. In Experiment II, the 2 x 2 x 2 design varied Importance, Confirmation, and Solution. For one solution, a significant Importance x Confirmation interaction was demonstrated; Ss in the Importance condition found disconfirmation unpleasant while those in the Unimportance condition found it pleasant. Results are interpreted in relation to dissonance theory and need for variety.