Although past research has offered important insights into how people seek to maintain their moral standing, it has generally portrayed this process as a matter of aggregating essentially static interpretations of a target's discrete acts. The present research reveals, however, that such interpretations are often far from static, and that they can change more than targets realize as new events unfold. More specifically, we find that: a) people can discount the diagnostic value of a target's initial deed if that party commits a subsequent act of the opposite valence, b) this occurs when an initial good deed is followed by a bad deed but not when the order is reversed, c) this occurs when evaluating the actions of others but not when evaluating the self, and d) this actor vs. observer difference can ultimately produce divergent beliefs about the target's overall morality, trustworthiness and subsequent trusting behaviors. We also identify a key mediating mechanism for these effects (i.e., the retrospective imputation of nefarious intent). Implications for reputation management, as well as the maintenance and repair of trust, are discussed.