The consequences of information system failure become more acute as organizations continue to invest in information technology and application development. Being able to better predict IS failure before implementation of a system could facilitate changes in the information system that can lead to implementation success. The realism of user expectations has been suggested as one possible means of assessing the eventual success or failure of an IS. Cognitive dissonance theory was used to hypothesize the behavior and attitudes of end users having certain expectations of a system. This experiment investigates the association between unrealistic expectations with both users’ perceptions (i.e., user satisfaction) and their performance with the IS (i.e., decision performance). A longitudinal experiment was performed in which the expectations of the subjects were manipulated to be unrealistically high, realistically moderate, or unrealistically low. The results suggest an association between realism of users’ expectations and their perceptions but not their actual performance. Future research should be directed toward the development of an instrument to measure user expectations, as well as toward understanding the causes of unrealistic user expectations.