This paper aims to show how the translator can benefit from developments in the availability of various sources for electronic text corpora, including CD-ROMS, on-line resources and the World Wide Web. The author argues that such resources yield up many texts of different types which the translator can then process automatically in order to identify what he calls a ‘natural’ translation. Illustrating how various types of search, using key lexical words and what might be called ‘linking’ or ‘pointer’ words can produce valuable terminological ‘hits’ for the translator, he suggests that paper dictionaries, both general and specialized, are not only less accessible from the user’s point of view but also less informative. The paper concludes that electronic corpora may, in fact, be of more practical use for translators than machine translation systems.