The use of meta-analyses has become increasingly widespread in flipped classroom research. They are typically seen as a more objective and credible method of summarizing the effects of the flipped classroom approach. However, problems among meta-analyses are long-standing and widespread, and can undermine the credibility of the results. This paper has two major objectives: (a) to examine the methodological features of the meta-analyses, and (b) to discuss the key methodological concerns of the meta-analyses. Nineteen flipped classroom meta-analyses were analyzed in this study, which included more than 495 unique primary studies, involving more than 72,200 flipped and 77,100 non-flipped participants. Given that at least 19 meta-analyses of the flipped classroom approach have been conducted so far, a critical appraisal of these meta-analyses is both important and timely because it will allow practitioners to evaluate the trustworthiness of the findings. The results of these meta-analyses suggest the mean effect sizes for cognitive outcomes (0.19–1.13), behavioral outcomes (1.40–3.12), and perceptual outcomes (0.05–1.62) vary from weak to strong support of flipped learning. However, some key methodological elements – searching for reference lists and grey literature, assessing the risk of bias in primary studies, non-independence of effect sizes, control for student initial differences, control for instructor equivalence, and assessing publication bias – were inadequately conducted or reported in the flipped classroom meta-analyses. There is positive empirical evidence for the flipped classroom, but there is also reason to question how this evidence has been synthesized. Copyright © 2021 Elsevier Ltd. All rights reserved.