Despite the many pedagogical benefits of self-/peer-assessment, they are not often practiced in the classroom, due to the fact that most teachers doubt learners' ability to do self-/peer-assessment accurately. Although several factors have been identified to affect self-/peer-assessment accuracy, the literature shows the rating accuracy of learners can improve if enough training is provided. Given the abovementioned supporting literature, it was hypothesized that learners, if provided with training and practice, may also have the potential to show behavior similar to that of expert-raters in terms of holistic and analytic intra-reliability. To test this hypothesis, having been trained to do self-/peer-assessment according to their group assignment, 136 English-major students conducted self-/peer-assessment of writing performance both holistically and analytically across 11 sessions. After correlating the students and raters' holistic and analytic scores and examining the variations among the correlations, it was found that students have indeed got the potential to show rating behaviors similar to those of expert raters and at times even show higher correlations. This paper closes with some implications these findings can have for theory and practice, and some new lines of research are recommended in the area investigated in this study.