Situated in the propositions of the Sociocultural Theory (SCT) about the role of social interactions in second language (L2) learning, the present study attempts to investigate the effect of different proficiency pairings on the Iranian EFL learners' academic writing. To do so, 40 male and female students of English literature/translation at the Department of English Language and Literature of Arak University participated in this study. The study was conducted during three sixteen-week semesters. All students had enrolled in the Advanced Writing course. The Oxford Placement Test (OPT) - 2nd version – was administered to establish the participants' level of language proficiency. All students also took a free-topic paragraph writing pretest. Based on the OPT test results, the students were paired with their peers at the same or different proficiency levels. Consequently, two types of pairings as the Same and the Mixed were made. The students received instructions on six types of academic paragraphs for 16 weeks. At the beginning of the semester, they were familiarized with the steps and procedures of providing and receiving peer feedback and they provided feedback on content, organization, grammar, mechanics and vocabulary of their peer's written paragraphs. The students took a posttest at the end of each semester. The pretests and posttests were rated by the raters. Additionally, the frequency of the feedback exchanged between the students was counted for each type of the pairings. The scores were compared using Pearson product moment correlation coefficient statistics, one-way ANCOVA, and independent sample T-tests. The results indicated that the mixed proficiency group had a significantly better performance in academic writing than the same proficiency group. Finally, the frequency of the peer feedback in both the same and the mixed proficiency groups was investigated. Possible reasons for the results and implications of the findings for second language teaching ar
