Recent ESP research into academic writing has shown how writers convey their stance and interact with readers across different disciplines. However, little research has been carried out into the disciplinary writing practices of the pure mathematics academic community from an ESP genre analysis perspective. This study begins to address this gap by applying Hyland's stance and engagement framework to pure mathematics research articles. The data consists of a corpus of 25 articles collected from five authors and semi-structured interviews with the same authors. The results of the corpus analysis reveal a low number of hedges and attitude markers compared to other hard and soft disciplines, but higher than expected shared knowledge and reader references. Furthermore, triangulation with interview data suggests that the epistemology and research practices of the discourse community can account for these frequency patterns, and that writers are conscious of the need to situate oneself within the norms of the discourse community by adhering to disciplinary writing conventions. The study also confirms that Hyland's framework can be usefully applied to pure mathematics research articles, although the boundaries between categories in the taxonomy are fuzzier than a stance/engagement dichotomy might suggest. (Contains 3 tables and 1 figure.)