Misinformation around scientific issues is rampant on social media platforms, raising concerns among educators and science communicators. A variety of approaches have been explored to confront this growing threat to science literacy. For example, refutations have been used both proactively as warning labels and in attempts to inoculate against misconceptions, and retroactively to debunk misconceptions and rebut science denialism. Refutations have been used by policy makers and scientists when communicating with the general public, yet little is known about their effectiveness or consequences. Given the interest in refutational approaches, we conducted a comprehensive, pre-registered meta-analysis comparing the effect of refutation texts to non-refutation texts on individuals’ misconceptions about scientific information. We selected 71 articles (53 published and 18 unpublished) that described 76 studies, 111 samples, and 294 effect sizes. We also examined 26 moderators. Overall, our findings show a consistent and statistically significant advantage of refutation texts over non-refutation texts in controlled experiments confronting scientific misconceptions. We also found that moderators neither enhanced nor diminished the impact of the refutation texts. We discuss the implications of using refutations in formal and informal science learning contexts and in science communications from three theoretical perspectives.