The article presents a methodology of pragma-communicative analysis of archival texts from the UN Parallel Corpus (documents from 1990 to 2014, translated into the six official UN languages), aimed at creating the principles of genre annotation of specialised corpora. Among the well-developed types of corpus annotation (prosodic, part-of-speech tagging, semantic, syntactic, etc.), genre annotation is particularly difficult, both from the perspective of linguistics and the perspective of creating automated computer programmes. In this case, pragma-communicative analysis can act as an effective linguistic tool that allows describing the genre composition of the studied corpus of texts, as well as conducting its functional analysis. The key terminological areas of the UN Parallel Corpus are defined as: «Законодательство/Legislation», «Юридические документы/Legal Documents», «Аналитические материалы/Analytical materials», «Деловая корреспонденция/Business Correspondence». The article contains analysis of such basic pragma-communicative functions of the studied corpus of texts as argumentative (represented by all types of recommendation documents: policy recommendations, working paper (recommendations with guidance), recommendation, note, public speech, argumentative part of a report), informative (represented by such genres as informative report, outline of schedule, programme of meetings, provisional overview, agenda, back-ground paper, executive summary), instructional (objectified by the following genre samples: administrative instructions, technical instructions, working paper, guidelines, decision, proposal, technical regulation), analytical (analytical part of White Paper, analytical report\ review, back-ground, injury, legal analysis of legislation, legal research paper, legal study) and regulatory (article, regulation, statute, convention, charter, charter, resolution, Bill, court statement, decree, norms, draft, special provision). A certain inventory of intratextual properties of corpus texts is described, including their lexical and morphological-syntactic characteristics. The article is concluded with possibilities of applying the proposed methodology to genre annotation of various types of corpuses.