<p>Technology-supported error-correction teaching (TS-ECT) is a key instructional practice in the field of STEM education, yet its effectiveness has not been systematically validated. Previous studies have reported mixed findings regarding the impact of TS-ECT on students’ STEM learning outcomes. Therefore, this study employed a three-level meta-analysis incorporating 37 experimental and quasi-experimental studies to examine the overall effect of TS-ECT on student learning and to identify potential moderators. The results indicate a medium-to-large overall effect size (<i>g</i> = 0.68), suggesting that TS-ECT facilitates achievement retention, error change, and students’ motivational-affective outcomes, whereas its effect on transfer performance is nonsignificant. Moderator analyses revealed that cultural context significantly influenced TS-ECT effectiveness, with collectivist countries showing substantially larger effect sizes than individualist countries. Discipline also played a significant role: the strongest effects were observed in science education, while effects in mathematics were nonsignificant. Regarding technology types, media resources and integrated platforms produced the most pronounced benefits, whereas dedicated teaching platforms yielded weaker effects. Furthermore, teacher guidance significantly strengthened instructional outcomes, whereas school level and learning mode exerted no significant moderating influence. Drawing on these findings, this study offers educational and technological implications to inform the future design and implementation of TS-ECT in STEM education.</p>