The use of corpora-collections of written and spoken texts stored on a computer-is becoming an increasingly influential procedure in the promotion of language learning, and nowhere is this influence more apparent than in English for Specific Purposes (ESP). Three main uses of corpora in ESP can be identified. First, in conducting research that underpins effective language teaching and learning. Second, corpora can be used as a way of presenting authentic language data to learners in an approach referred to as 'data-driven learning' (DDL). Third, corpora can be a resource for teachers to use in preparing classroom teaching activities and materials. This paper evaluates the use of corpora in each of these areas, highlighting not only its considerable potential in providing more accurate information about language in use than is sometimes included in textbooks and reference works, but also the challenges that face researchers, teachers and materials writers in employing this information to develop effective classroom practice.