Background: Self-related information is difficult to ignore and forget, which brings valuable implications for educational practice. Self-referential encoding techniques involve integrating self-referencing cues during the processing of learning material. However, the evidence base and effective implementation boundaries for these techniques in teaching and learning remain uncertain due to research variability. Aims: The present meta-analysis aims to quantitatively synthesize the results from studies applying self-referential encoding techniques in education. Methods: The analysis was based on data from 20 independent samples, including 1082 students from 13 primary studies identified through a systematic literature search. Results: Results from random effect models show that incorporating self-referential encoding techniques improved learning (g = 0.40, 95% CI [0.18, 0.62]). Subgroup analysis showed that the valence of learning material serves as a significant boundary condition for this strategy. The students' cohorts, types of learning materials, and research context did not moderate the effect sizes. Conclusions: Our results suggest that incorporating self-referential encoding techniques on negative materials shows an aversive effect. Overall, there is a universal benefit to using self-referential encoding techniques as an appropriate design guideline in educational contexts. Implications for teaching practice and future directions are discussed. Further studies are needed to investigate the effectiveness in more diverse educational and teaching situations.