Retrieval failures--moments when a memory will not come to mind--are a universal human experience. Yet many laypeople believe human memory is a reliable storage system in which a stored memory should be accessible. I predicted that people would see retrieval failures as aberrations and predict that fewer retrieval failures would happen in the future. After responding to a set of trivia questions, participants were asked whether they would do better, about the same, or worse if they were given a different, but equally difficult, set of questions to answer. The majority of participants said they would do about the same, but more participants said they would do better next time than said they would do worse, although these participants did not actually do better. This finding was especially pronounced when participants were given feedback, suggesting that hindsight bias--the feeling, which emerges when an answer is presented, that one knew it all along--contributed to participants' belief that they had underperformed on the first set of questions. The finding that metacognitive judgments were influenced by beliefs stands out in a literature full of studies in which beliefs fail to influence judgments.