The aim of this comparative study is to investigate the deployment of interactional metadiscourse features in two different academic genres. For this purpose, a small, specialized corpus of 48 research articles and book reviews from seven different disciplines were collected. The conclusion sections of the texts written by non-native speakers of English were investigated to find out how interactional metadiscourse features were used. Drawing on previous metadiscourse frameworks, "hedges," "boosters," "attitude markers," "self-mentions" and "engagement markers" were identified in both sub-corpora. The results indicated significant differences across the groups on how writers constructed their authorial stance with interactional metadiscourse markers. Findings revealed that by using a rich number and variety of attitude markers, book reviewers were more evaluative in their conclusions. On the contrary, higher use of hedges in research articles allowed the authors sound more cautious in their commitment to the propositions. This study offers a detailed account of interactional metadiscourse in these two genres and illustrates how interpersonal function of language is accomplished for particular purposes in different academic texts.