The analysis of domain-specific terminology is essential for characterizing specialized discourses, and emerges as a useful means of measuring the thematic hybridity of law and legal translation in particular. This paper accordingly presents a large-scale mapping of terminological and phraseological features in a multi-genre corpus that was built as part of the LETRINT project on institutional legal translation. The corpus-driven analysis focuses on the density of legal terminology and phraseology, on the one hand, and that of terminology of other specialized domains, on the other, in nine genres that are considered representative of three central legal functions (law-making, compliance monitoring and adjudication) in three international settings (the European Union, the United Nations and the World Trade Organization). The comparative examination of density scores provides empirical evidence of the common core features of the selected genres, and reveals variations based on institutional thematic focus, primary legal function and genre specificities. These insights nuance our understanding of international legal discourses and domain specialization in institutional translation.