In this study, we examined monitoring accuracy during in class-exams for Emirati, American and Cypriot college students. In experiment 1, 120 students made local, confidence-ratings for each multiple-choice question in a psychology exam and also estimated their performance at the end of the exam. In experiment 2, to investigate the effect of practice in monitoring accuracy, 69 students made confidence judgments in two consecutive exams. In Experiment 3, to evaluate whether rating one's confidence aids exam performance, 172 students completed the exam either with or without providing confidence judgments. In all three cultures, confidence judgments were accurate indicators of performance, but there were significant cross-cultural differences that probably arose from regional educational practices. Additionally, global judgments were subjected to bias. Whereas American students overestimated their performance, Emiratis underestimated it. Practice alone with confidence judgments did little to improve monitoring accuracy. But there was a significant benefit for students who rated their confidence compared to those that did not. The results are discussed in light of the literature on metacognitive accuracy.