We're not exactly taking machetes to the reference collection in my library, but we definitely are thinning the herd, as it were.
What we're finding, simply put, is that while students use and love online journal articles and, increasingly, electronic books (shortly after we highlighted our NetLibrary collection, I got two reference queries about how to get e-books specifically), the reference collection is mostly gathering dust.
This problem isn't unique to us, by any stretch. But when you're out of shelf space and have a ten-year-old encyclopedia in front of you that's only been used once since you bought it, you have to ask yourself: do we have to keep this?
There's a disciplinary angle here. Currently I'm weeding the business reference collection, and business is a discipline where currency is important--not money, though that's important too, but the timeliness of the material. I also work in the sciences, which has the same emphasis on currency, and in nursing, where accreditation requirements have led to tossing a great deal of old material. In music, in contrast, keeping those old materials might be the only way of researching something ancient and/or obscure. We once received a call from a faculty member at another university because we were the only library in the country with a copy of a particular score. Stuff like that is why I get up in the morning.
Reference books, though, don't go through Interlibrary Loan; not only because they're, well, reference, but because the information needed tends to brook few delays like that required to get an ILL book to a patron. Which in turn explains the growing reliance on Google and Wikipedia, to cite the two best-known examples. I'm not sure why the convenience of these tools is so often decried in the library community; yes, people should make sure that the information they're getting is good, but shouldn't the resources that contain that good information make themselves more accessible? I used to maintain a personal subscription to Britannica Online, because--get this--it was more convenient than going downtown to use the print encyclopedia in the library. I think I gave up when they neglected to send me a renewal notice, and I had to inquire (and wait a few days for a response) to find out why it wasn't working. Perhaps they've improved.
The need for the kind of information that usually gets classed as "reference" isn't going away; the massive volume of ready-reference-style searches on free Web search engines is evidence enough of that.
The need for reference books, though? That's an open question.