Michael Barlow's corpus linguistics site is a good starting point and has lots of links to sites where you can find both English, and foreign language corpora.
The Lancaster University based UCREL (University Centre for Computer Corpus Research on Language) site gives information about a number of corpus-based projects, including the ATR, the BNC, CRATER, MARSEC and COLT.
The Collins Cobuild site allows you to carry out a collocation demo of their 50 million word corpus.
The ICAME site (International Computer Archive of Modern and Medieval English) contains among other things a small sample of the Brown corpus.
This site gives information about the corpora mailing list as well as information about a range of corpora.
The Tim Johns Data-driven learning page is a Birmingham University based site, and is especially concerned with the use of corpora in teaching.