The role of counterfactual thinking in two specific emotions, shame and guilt, was examined in two series of studies. In Studies 1a and 1b subjects imagined themselves in two situations, one evocative of shame, and the other of guilt. Subjects generated counterfactual alternatives in order to "undo" the distressing outcomes. Consistent with predictions derived from Tangney (1991), subjects tended to undo the shame situation by altering aspects of themselves (personal qualities, attitudes), and to undo the guilt situation by altering actions. In Study 1c subjects described autobiographical experiences in which they had felt shame or guilt and provided counterfactuals to undo the experiences. Findings paralleled those of the previous studies. In a second series of studies, subjects imagined themselves in a situation that could elicit either shame or guilt. In Study 2a half of the subjects were led to mutate personal qualities, and the remainder were led to mutate actions to undo the situation; all subjects then rated their anticipated feelings of shame or guilt. The mutate-self subjects anticipated greater shame whereas the mutate-behavior subjects anticipated greater guilt over the outcome of the situation. In Study 2b, one-third of the subjects engaged in no counterfactual thinking, while the others mutated the self or behavior to undo the outcome of the situation. Compared to the no-mutate condition, mutation manipulations had amplifying effects on shame and guilt. Discussion centers on the role of counterfactual thinking in shaping specific emotions and in differentiating shame- and guilt-prone personalities.