This chapter discusses the current role of natural language processing in lexicography, and considers how this might change in the future. It first considers the shared history of natural language processing and lexicography with respect to statistical methods. It then discusses how natural language processing is applied to pre-process corpora to support lexicographic analysis, identify collocations in corpora, automatically construct thesauri, and select good dictionary examples. It also discusses the natural language processing tasks of word sense disambiguation and induction and their relationship to lexicography, and very recent neural network-based methods for automatically generating definitions. It concludes by discussing specialised types of dictionaries that can currently be automatically constructed, and considers whether dictionary construction could ever be fully automated.