Rumination, a stable vulnerability marker of depression, is associated with lack of affective bias towards positive information due to cognitive control deficit. However, whether ruminators are self-aware remains unknown. Based on the signal-detection approach, the present study examined metacognition associated with rumination. Specifically, metacognitive sensitivity, the extent to which one’s confidence tracks task performance, was measured using the meta-d’ method. In three directed-forgetting memory experiments, the present research examined whether trait rumination (Experiment 1) and interaction between trait rumination and mood state (Experiments 2 and 3) influenced metacognitive sensitivity. Participants were asked to either forget or remember some affective pictures (i.e., pictures with negative or positive valence). They then performed an old-new discrimination task and indicated confidence rating for their responses. In Experiment 1, an interaction between Valence and Rumination was found: High ruminators’ meta-d’ level decreased significantly when attending to positive-valence images. In Experiments 2, participants performed a similar memory task following either sad- or neutral-mood induction. An interaction among Valence, Rumination, and Mood was observed: During sad mood, high ruminators’ meta-d’ level for processing positive-valence images became significantly lower relative to low ruminators. In Experiment 3, participants’ memory task was preceded by either a rumination- or distraction-focused task. Similarly, an interaction among Valence, Rumination, and Task indicated lower meta-d’ level when high ruminators attended to positive-valence images after rumination task. These findings suggest that trait rumination and sad mood state are associated with metacognitive processing deficits in positive information. Clinical implications are discussed.