This lecture considers what reference and pedagogical grammars and grammar teaching materials for L2 learners should ideally include, based on corpus evidence from both native-speaker and learner corpora. I demonstrate how learner corpora can be used to track the emergence of grammatical features, from the elementary level to advanced, how learners get to grips with new grammar and what we can learn from the statistical output of error-coded corpora. Additionally, we look at how the divide between lexis and grammar becomes progressively blurred and how corpus information can best be used to produce useful grammars and teaching materials for students at different levels. The advanced level in particular is focused on. The lecture is presented within the framework of the English Profile Project (EPP), a large, international, inter-disciplinary project which uses corpora to investigate learner competence at different levels of the Common European Framework of Reference (CEFR).