This first study intends to contribute to the analysis of referential cohesion by revising part of Halliday and Hasan’s (1976) slightly outdated but still frequently cited and used theory-based taxonomy of cohesion. This chapter describes the process of developing and piloting a new analytical tool for Referential Cohesion Analysis. Modifications were made with the intention of improving its applicability for the analysis of larger corpora of texts in a more reliable way. The process is supported by an exploratory analysis of a corpus of 10 research articles (RAs) to reveal what information about cohesive referential patterns in research articles Referential Cohesion Analysis yields – especially as compared to Halliday and Hasan’s (1976) approach – and to summarize the emergent hypotheses and further research questions that have been raised throughout the analytical process. Results from the analysis of the 10 RAs show that the analytical tool can capture patterns in referential cohesion that have not been accounted for by Halliday and Hasan’s (1976) taxonomy, for example, chains of referential cohesive ties. As a result of the clear visual representation of cohesive chains, referential links become easily searchable; furthermore, some calculations can be automated even in a large corpus, which hopefully enhances the reliability of the analysis.