As past studies showed mixed results, this meta-analysis determined game-based learning's overall effect on students' critical thinking and tested for moderators, using 21 effect sizes from 20 empirical studies of 1,947 participants. The results showed that game-based learning had a significant positive overall effect on students' critical thinking (g = 0.863, k = 21) and showed significant heterogeneity among effect sizes. Among game types, role-playing games yielded the largest mean effect size (g = 1.828, k = 5). The effect size of game-based learning was larger for critical thinking disposition (g = 1.774, k = 4) than critical thinking skill (g = 0.661, k = 17). Game-based learning also had a larger effect on students in collectivistic countries (g = 1.282, k = 10) than those in individualistic countries (g = 0.432, k = 10). Furthermore, this effect size was larger in later publication years. Lastly, the effect size of game-based learning on critical thinking was larger for studies published in journal articles (g = 1.154, k = 13) than theses (g = 0.378, k = 8).