Contemporary theories of metacognitive monitoring propose that beliefs play a critical role in monitoring of learning. Even so, recent evidence suggests that beliefs are not always sufficient to impact people's monitoring. In seven experiments, we explored people's beliefs about the impact of mood and item valence on memory and whether people use their beliefs about these cues when monitoring their learning. Participants expressed mood-congruent beliefs (Experiments 1, 6, and 7). That is, they believed people in a negative mood would remember more negative items than positive and neutral items. To evaluate whether they use this belief when monitoring their learning, participants studied emotional (positive and negative) and neutral pictures (Experiments 2 and 3) or words (Experiments 4, 5, and 7), made a judgment of learning (JOL) for each, and completed a free-recall test. In Experiments 2--5, participants completed the learning task while in a negative or neutral mood. The negative mood was induced with an established mood induction procedure. In contrast to the belief-based hypothesis, participants did not make mood-congruent JOLs; JOLs were not influenced by mood. By contrast, JOLs were consistently higher for emotional relative to neutral items. Thus, although participants demonstrated a mood-congruent belief, they did not use this belief when monitoring their learning. These outcomes demonstrate that simply having a belief about a person-centered cue (e.g., a belief about the impact of a person's mood on memory) is not sufficient for that belief to impact monitoring of learning.