An increasing number of people involved in medicine are under pressure to publish research, but there is little understanding of how to describe structured writing.This paper aims to describe the structure of original research papers published in the British Journal of General Practice with a view to providing insight into the nature of such analyses, and particularly to help researchers and trainers to write and teach writing more successfully.A sample of 50 original papers published in the Journal between January 1989 and March 1993 were examined. The papers were subjected to a form of 'move structure analysis', a technique used in applied linguistics; move structure analysis assigns a tentative function to a piece of text, and identifies words/phrases associated with it. To be recognized, moves thus identified had to occur in the same section of the paper in 65% of the corpus, and/or appear in the same order relative to other moves in 50%.Fifteen moves were identified, four in the introduction, three in the method, and four each in the results and discussion. These moves functioned, for example in the discussion, to state limitations and defend successes; describe achievements; contextualize procedures and findings; and offer recommendations. Frequency scores ranged from 66% to 100%, and order stability scores from 50% to 80%, with three moves being unordered.It is possible to derive from this study a template for structuring academic medical writing. This template may be built up from the exemplary quotations in the text, to provide assistance to educators and less experienced writers.