We have a fair understanding of what contributes to our confidence when performing individual trials of a task. However, little is known regarding the factors driving more global metacognitive estimates when a task is repeated. The present study investigates the contribution of uncertainty to global performance monitoring. In two pre-registered experiments, participants performed four trials of an orientation matching task and reported their mean response and an estimated dispersion around this perceived mean as a proxy for global performance monitoring. We considered several sources of uncertainty: response-related uncertainty, related to the participants and observed in their response variability, and perceptual or attentional uncertainty related to the sensory stimulation. Our results suggest that adults can reliably estimate the mean and dispersion of their performance and use it together with stimulus-dependent uncertainty to inform their global performance monitoring. In particular, participants adequately report that their performance was worse when uncertainty was higher. However, this capacity decreases when different types of uncertainty increase jointly. We discuss these results in light of a model of confidence that reproduced our main findings. These behavioral and computational results clarify the role of uncertainty in perceptual metacognition and the relationship between local and global performance monitoring.