The present study examined peer feedback accuracy in synchronous and asynchronous peer feedback conditions. Employing a counterbalanced repeated-measures design, the researchers examined the accuracy of 2327 comments provided by 96 intermediate EFL (English as a Foreign Language) learners in synchronous and asynchronous computer-mediated conditions. In the synchronous condition, the learners exchanged comments using the desktop version of WhatsApp, and the participants in the asynchronous condition created weblogs and provided feedback in the comment box. All students provided peer feedback in both conditions, and the accuracy of their comments in these computer-mediated environments was compared. Findings showed that the asynchronous condition yielded significantly more accurate comments and fewer missed erroneous items. The data analysis also indicated that low-intermediate students provided more accurate comments in the asynchronous condition. Further, the findings revealed that EFL learners provided comments supported by references to external sources, including textbooks, online sources, teachers' lectures (referentially-justified peer feedback), which were significantly more accurate than those comments not accompanied by references. The findings of this study suggest that when the accuracy of peer feedback is of paramount importance, asynchronous peer feedback design should be implemented.