Research on innovation adoption has suffered from a bias towards understanding the factors that affect the dichotomous adoption/non-adoption decision.Much less attention is devoted to the question why potential adopters fail to progress to the adoption stage from earlier stages in the decision makingprocess. Such knowledge is essential to understand what factors actually underlie the non-adoption of an innovation. As perceived innovationcharacteristics have been found to influence adoption in a substantial way, we develop hypotheses on their influence not only with respect to theadoption stage, but with respect to previous stages of the adoption process as well. Specifically, we develop hypotheses on the perceived levels andimportance of relative advantage, compatibility, complexity and perceived risk in the awareness, evaluation, and adoption stages of the innovationadoption process. The hypotheses are tested using both multivariate analysis of variance and multinomial logit modeling on a sample of 242organizations, focusing on the adoption process of electronic banking. The results show that the levels of perceived relative advantage andcompatibility increase over the stages of the adoption process, whereas the perceived levels of complexity and risk largely decrease. The influence ofthe characteristics across the adoption stages shows that positive beliefs related to the innovation have highest salience in the initial stage of theprocess, whereas the salience of perceived complexity - generally considered an undesirable attribute - is highest in the final stage. In sum, ourresults imply that non-adopters are affected by innovation characteristics in a different way, depending on their stage in the adoption process, andtherefore should not be considered as one homogeneous group of "potential adopters". These findings have important implications for marketinginnovations.