Research shows that people display a downward shift in their predictions in anticipation of performance and feedback. The authors used a misattribution paradigm to explore whether anxiety serves as a signal for predictions. Participants (N = 108) anticipating results from an important test either immediately or in a few days were or were not encouraged to attribute any arousal they experienced to coffee they consumed earlier. Consistent with predictions, participants encouraged to attribute their arousal to the coffee were optimistic in their predictions even when anticipating immediate test feedback. In addition, the more participants attributed their arousal to the coffee, the more optimistic they were in their predictions. Ancillary analyses suggest that anxiety can be a cause rather than a consequence of less optimistic predictions.