Educators often use stories, humor, surprise, images, or other ploys to catch students' interest. This form of interest, often called "situational interest" because of how it is triggered by contextual cues, can facilitate learning to some degree. Yet it also may carry risks. This article proposes one such potential risk: flawed metacognition, in the form of inflated judgments of learning (JOLs) and overconfident estimates of future performance (calibration bias). Two experiments tested this hypothesis. In each, college students (Ns = 201, 196) read passages on a novel topic (the physics of lightning in Study 1; the Hare Krishna organization in Study 2), crafted to be relatively dull or interesting. They then reported their interest, JOLs, and performance estimates before completing a test of their topic knowledge. Otherwise, their methods differed in several ways, such as how situational interest was induced and how performance was measured. Despite those differences, the findings from each study confirmed that situational interest inflated participants' JOLs and promoted overconfident performance estimates. Furthermore, Study 2 showed that those inaccurate metacognitive judgments affect studying decisions; due to their inflated JOLs, participants allocated less of their limited time for restudy to the fun passages. Potential moderators are considered, as are implications for interest theory, for metacognition theory, and for educators more generally.