Peer feedback has been widely used in English as a foreign language (EFL) writing classes, but its effect is highly variable and its quality and validity are often questioned by students as well as teachers. This research aims to explore the effects of teacher intervention on student use of different types of peer feedback and on their self-revision of writing texts. The study employed a quasi-experimental design with two experimental groups receiving either the teacher's direct correction for erroneous peer feedback or indication of inaccurate peer feedback with symbols and a control group receiving no teacher intervention on peer feedback. One hundred and ten non-English major graduate students in three parallel writing classes participated in the study. The study lasted for one and a half months, during which the participants completed two essays, conducted peer feedback, and revised their essay drafts based on the peer feedback received. The adoption rates and accurate revision rates of different types of peer feedback as well as self-correction rates for the three groups were compared and analysed. It was found that the two groups with teacher intervention made significantly more accurate revisions in the subsequent revised essays higher than the control group. Additionally, the adoption rate and accurate revision rate of direct correction were significantly higher than the other two types of feedback. These results have implications for incorporating peer feedback in EFL writing class and for future peer feedback research.