This work examines how people form evaluations of extended experiences that vary in valence and intensity. It is documented that when people retrospectively evaluate such experiences, not all information is weighted equally. Some prior research demonstrates that earlier parts are weighted more than later parts, while other research shows the opposite. In this paper we suggest that differences in evaluation tasks shift the focus to different aspects of the experience, causing individuals to be differentially influenced by earlier or later parts of the experience. We show that ratings of feelings (hedonic evaluation tasks) lead to stronger preferences for improving experiences than do evaluative judgments (informational evaluation tasks), suggesting that later aspects of the experience are weighted more heavily in affective tasks. In addition, we investigate other evaluation tasks, demonstrating that whether the task is descriptive or predictive and whether the target of the evaluation is the source of the experience or the experience itself also alter the weight given to different parts of the experience. Our studies demonstrate systematic shifts driven by these different evaluation task, revealing changes in overall evaluations as well as changes in the underlying weighting of key characteristics of the experience (i.e., start, end, and trend). Copyright © 2006 John Wiley & Sons, Ltd.