The use in translation studies of methodologies inspired by corpus linguistics has proved to be one of the most important gate-openers to progress in the discipline since Toury's re-thinking of the concept of equivalence. In this paper, the author discusses mainly the use of corpora in translator education, arguing that, the many advantages to be had by this method notwithstanding, there are problems. For example, it is not always obvious which corpus might help a translator solve a specific problem; corpus evidence might be misleading in some cases; and offering past linguistic behaviour as a model for future flies in the face of the nature of language and may, furthermore, stifle invention. For these reasons, the author argues, it is worth exploring ways of using corpora which may seem subversive of standard uses, either alone or in conjunction with more traditional methods of investigation and teaching in Translation Studies.