Refutation texts are beneficial for removing misconceptions and supporting comprehension in science. Whether these beneficial effects hold true in the domain of statistics is, however, an open question. Moreover, the role of refutation texts for the accuracy in judging one’s own comprehension (metacomprehension accuracy) has received little attention. Therefore, we conducted an experiment in which students with varying levels of statistical misconceptions read either a standard text or a refutation text in statistics, judged their text comprehension, and completed a comprehension test. The results showed that when students read the standard text, having more misconceptions resulted in poorer text comprehension and more inaccurate metacomprehension as indicated by overconfident predictions. In contrast, when students read the refutation text, the number of misconceptions was unrelated to text comprehension and metacomprehension accuracy. Apparently, refutation texts help students to pay attention to inaccuracies in their knowledge and, thereby, can promote self-regulated learning from texts.