I compressed some random HTML files tonight, and just for the sake of it, compared zip, gzip and bzip2. The results turned out to be quite interesting.
The data set consisted of about 140 rather similar HTML files, taking up 27.713.253 bytes in total. These are the file sizes in bytes of the final archives:
Or in other words, zip and gzip compressed the data down to about 5.3% of its original size, which is rather good. However, bzip2 did almost four times better, and used only 1.5% of the original size!