Digital game-based learning (DGBL) interventions can be superior to traditional instruction methods for learning, but previous meta-analyses covered a huge period and included a variety of different target groups, limiting the results' transfer on specific target groups. Therefore, the aim of this meta-analysis is a theory-based examination of DGBL interventions' effects on different learning outcomes (cognitive, metacognitive, affective-motivational) in the school context, using studies published between 2015 and 2020 and meta-analytic techniques (including moderator analyses) to examine the effectiveness of DGBL interventions compared to traditional instruction methods. Results from random-effects models revealed a significant medium effect for overall learning (g = 0.54) and cognitive learning outcomes (g = 0.67). Also found were a small effect for affective-motivational learning outcomes (g = 0.32) and no significant effect for metacognitive learning outcomes. Additionally, there was no evidence of publication bias. Further meta-regression models did not reveal evidence of moderating personal, environmental, or confounding factors. The findings partially support the positive impact of DGBL interventions in school, and the study addresses its practical implications.