Rosenberg [1976] argues that in many markets prospective buyers for an innovation are strongly influenced by expectations "...concerning the timing and significance of future improvements..." The primary objectives of this article are to provide a model of the innovation decision process of user firms that expect improvements in current best technology and to offer an empirical study which tests the model's predictions. Among other results, we find empirical support for Rosenberg's argument and demonstrate theoretically that, compared to models without technological expectations and learning, the optimal decision process is non-monotonic.