There has been increasing interest in the possible applications of corpora to both linguistic research and pedagogy. This study takes a corpus-based, genre-analytic approach to discipline-specific materials development. Combining corpus analysis with genre analysis makes it possible to develop teaching materials that are not only authentic but also research-supported. An RA corpus consisting of 60 research articles from three major journals of computer science was constructed. A word frequency list derived from the corpus was analyzed to develop a vocabulary profile for the genre. Move analysis was also conducted based on a self-developed coding scheme of rhetorical moves in the target genre. The move codes were tagged in the corpus texts so that individual moves and move patterns could be retrieved for the purpose of developing research-based online teaching materials for graduate students of computer science. Numerous examples of specialized vocabulary, grammatical usage, and move structures that showcase and characterize academic computer science discourse were used in not only lessons but also learning tasks, discussion topics, and online writing models. The paper ends with a discussion of the usefulness and effectiveness of the online RA writing materials, based on student feedback and assessments. (Contains 5 figures and 2 tables.)