This study analyses peer assessment through many facet Rasch model (MFRM). The research was performed with 91 undergraduate students and with lecturer teaching the course. The research data were collected with holistic rubric employed by 6 peers and the lecturer in rating the projects prepared by 85 students taking the course. This study analyses raters, measurements for students who are rated, criteria used in rating and extent to which rubrics fulfil their function. Moreover, it also investigates effects of peers' levels of achievement on the process. In consequence, it was found that raters differed in the levels of strictness and generosity in rating, and that students were distinguished adequately in terms of the property measured. Besides, a very high level of reliability value was estimated in relation to the criteria in the study. This was interpreted as that they functioned in a reliable way in distinguishing between students' performances. It was found in the analyses of achievement levels of peers taking part in peer assessment that ratings made by students with high levels of achievement differed significantly from those made by students with medium or low level of achievement. Finally, the views about peer assessment were generally positive.