KeywordsPeer review process Referee report of scientifc articles Scientifc writing Discourse analysisThe referee report of research articles is a discursive genre in which a part of the collective process of building knowledge is crystalized. Given its confdentialityl there is litle empirical work on this genre. Among these few contributionsl a series of limitations can be identifedl such as the lack of defnition of the unit of analysis or the lack of integration between discursive and sociometric data. In this articlel we present a model that contributes to overcoming these limitations. The model was applied with the purpose of analyzing the variation of referee reports according to two extralinguistic variablesl i.e.l publication recommendation (acceptedl conditionedl rejected) and referee's number of published articles. The results showed that both extralinguistic atributes were associated with specifc discourse variationsl of which publication recommendation was more decisive than referee's academic productivity.