This paper presents the main lines of a genre analysis of the macro-structures of research articles (RAs) in applied linguistics, an area that deserves more attention both for pedagogic and research reasons. The analysis is based upon a detailed study of a corpus of 40 RAs, selected as random sets of 10 drawn from four leading journals in the mid-1990s. Half the corpus proved to consist of primary RAs of various kinds, while half comprised secondary research. Our findings for primary RAs in our corpus show how they both draw upon and vary from the conventional "Introduction--Method--Results--Discussion" (IMRD) framework. We describe the realizations of IMRD and other optional elements in our analyses, and examine some complexities resulting from differences in the use of section headings and in the apparent conceptualization of the function of some sections. For secondary RAs, largely neglected in previous research, we present an initial macro-structural account, i.e. Introduction--Argumentation--Conclusion, and briefly comment on the internal structuring of central Argumentation sections, which reflects differences in the overall communicative purposes of secondary RAs.