Academic writing, much like any other kind of writing, is only effective when writers use conventions that other members of their community find familiar and convincing. Essentially the process of writing involves creating a text that we assume the reader will recognise and expect, and the process of reading involves drawing on assumptions about what the writer is trying to do. It is this writer-reader coordination which enables the co-construction of coherence from a text. Scholars and students alike must therefore attempt to use conventions that other members of their discipline, whether journal editors and reviewers or subject specialist teachers and examiners, will recognise and accept. Because of this discourse analysis has become a central tool for identifying the specific language features of target groups. In this paper I draw on my own work, conducted over several years into research and student genres, to show how some familiar conventions of academic writing are used in different disciplines and what these differences can tell us about the work in the disciplines themselves.