Digital game-based learning (DGBL) is known to be widely used for improving learning in various fields. Among the elements of DGBL, competition has been very controversial. This meta-analysis, which included 25 articles written between 2008 and 2019, revealed that DGBL has produced improvements for learning outcomes with an overall effect size of 0.386. In addition, we explored multiple moderators to understand how competition in DGBL influenced student learning for different learners, contexts, game types, and learning outcomes. We found that competition in DGBL was effective for math, science and language, but not for social science and other subjects. It was effective for K12 students and college students. It was effective for puzzle, strategy, role-playing, and simulation, but not for action games. Finally, competition in DGBL was equally effective for cognitive and non-cognitive outcomes. Through the results of this study, we fill a critical gap in the research left by recent reviews, which do not examine the role of competition; a key gaming element. In addition, we offer a number of suggestions for future studies.