The meta-analysis investigates the impact of active learning on K--12 students' academic achievement and learning retention. A thorough search of published and unpublished research reveals 398 studies for academic achievement and 88 studies for learning retention. The analysis demonstrates that active learning has a significant positive impact on academic achievement (g = 1.005) and learning retention (g = 1.204) in comparison to teacher-led instruction under the random-effects model. The analysis conducted to identify publication bias suggest that the findings are not influenced by publication bias. However, there is significant variation in effect sizes across studies. Therefore, moderator analysis is conducted for eight possible moderator variables. The analysis indicate that the impact of active learning on academic achievement is moderated by domain subject, educational level, course implementer, region, and group size. Conversely, the impact of active learning on learning retention is moderated by publication type, domain subject, educational level, experimental duration, and region.