<p>Previous studies have shown that our metacognitive assessments, simply described as “knowledge about knowledge”, are not always accurate. In the current study, we examined people’s Judgments of Resolving Time (JoRTs) and compared them to their Actual Resolving Time (ARTs) for math and verbal tasks. Our aim was to test whether people tend to over- or underestimate their task completion times. We hypothesized that individuals would display overconfidence on verbal tasks, based on data from the hindsight bias or the “I knew it all along” findings, but perhaps underconfidence on math tasks—overestimating their JoRTs—given the prevalence of math anxiety. We also investigated how JoRTs might influence persistence and confidence in each domain. The results were surprising and complicated. First, people were underconfident on the verbal tasks but fairly accurate on the math tasks. Furthermore, the JoRTs were positively correlated with persistence in the math domain but negatively correlated with confidence in both domains. Finally, we found that confidence mediates the relationship between JoRTs and persistence in the math domain, and present the resulting path diagram.</p>