Peer feedback is a frequent feature of second language (L2) writing instruction. However, whether it is facilitative of L2 writing development is controversial. This meta-analysis aggregates the quantitative results of 26 empirical studies to (1) provide an accurate estimate of peer feedback's effectiveness in improving L2 writing, (2) compare its effects with those of other sources of feedback, and (3) identify factors moderating its effectiveness. Results revealed that peer feedback has a significant positive overall effect on students' writing: d = 0.73, 95% CI: 0.54-0.92. No significant differences were found between the effects of peer feedback, teacher feedback, and self-revisions; however, this finding is tentative given the small number of primary studies comparing peer feedback and feedback from other sources. Moderator analyses revealed larger effect sizes when students had more time to write and when treatments were longer. The analyses further showed that peer feedback had greater effects on revisions than new compositions and that peer feedback was more effective for content than language (lexis and grammar). Based on the findings and methodological features of the primary research, recommendations for further research are proposed so that firmer conclusions may be obtained in future meta-analyses.