This meta-analysis aims to examine the overall effect of a form of peer-assisted learning, peer tutoring, on students’ L2 writing performance and abilities across educational levels and contexts. A literature search via five databases (Scopus, ERIC, Education Source, British Education Index, and APA PsycINFO) retrieved 26 eligible experimental studies, yielding 190 individual effect sizes for meta-analysis. The meta-analysis suggested a large favourable overall effect size (Hedges’ g = 0.84, P < .001) of peer tutoring on L2 writing performance and abilities. The regression model showed that the integration of metacognitive prompts with peer tutoring will significantly increase the treatment effect. Given the significant large average positive treatment effect, L2 writing educators are recommended to adopt peer tutoring in their classrooms.