Last year I was involved in writing a large group proposal with LaTeX. This involved various sections written and rewritten by disparate subgroups, with all the merging and editing headaches you'd expect and more. In particular, handling the bibliography was a nightmare. Everyone handled references by hand without using BibTeX or anything similar (some people didn't even use \cite), and it took hours of work at the end to get a properly alphabetized bibliography without duplications; we ran out of time and energy for any attempt at consistency of formating, or even checking which preprints were now publications in print.
It is likely that I will be involved in a similar proposal at some point in the future, and I'm looking for suggestions to make (at least) the bibliography work more smoothly.
Keep in mind that what I'm dealing with is at least as much a sociological problem as a technological one. I think almost none of the people I'm working with have even heard of BibTeX, and some may be resistant to making the time investment to learn any unfamiliar software at all. So very low overhead is a must.
I had been planning to insist on using BibTeX, but I just recently learned about AMSRefs, which (used in the simplest way) seems to require less overhead for novices. (Since we are mathematicians, it's easy to get both BibTeX and AMSRefs data for our citations from MathSciNet or MRefs.) And of course there may be other alternatives, or things that work with BibTeX, that I don't know about.
I'm well aware that BibTeX or something similar would at best only help with a very few aspects of the problems I described with my past experience, and that the Right Solution would be to use some kind of VCS for the entire thing. Unfortunately, I think it's probably out of the question to get everyone else to learn to use such a system (though I'm open to being convinced otherwise).