There's a lot of negativity in this thread, which surprises me a little. I had a blast when I did my PhD, but it may be different in the humanities. Here's my story.
I did mine from Sep 2001-Sep 2004 in the Computer Science department of Cardiff Uni. Pay wise, I did pretty well. Standard stipend from the EPSRC (quango that pay for PhDs in the physical sciences) was £8.5k per year, and my PhD was part sponsored by BAE Systems[1] so I had another £3k per year on top of that. Now that's not loads, but it is tax free. Most PhD students also have a chance to help out with the teaching in the department, running tutorials and suchlike, and that pays pretty well, around £15 per hour. Because the stipend is tax free, that means the tutorial income is inside your income tax threshold, so there are no deductions. Upshot is that I was on the equivalent of about £22k per year, which was pretty nice, and a world away from living on student grants as an undergrad.
The main reason my PhD was good was my supervisor. When I started, he already had the problem mapped out, so I had no messing around; I was straight on it. In addition, the radio engineering group is quite big in Cardiff (3 staff, 5-7 postdocs, 4-5 doctoral candidates) so there were three of us starting at the same time. This meant it wasn't as lonely as it could have been -- we all worked in roughly the same area, close enough to understand each others work (this is rarer than you may think), so there was always some to talk problems over with. Having said that I spent most of year 3 and all of my subsequent work (like a lot of PhDs I overspilled into year 4) working at home to cut down on my travelling, so I think I had the loneliness thing as bad as anyone. It's manageable but it does suck sometimes.
So, yeah, supervisors.
Your supervisor choice is critical -- that link isn't a joke, it's a parable. Your supervisor needs to be able to make time for you, and willing to keep tabs on your work. They need to be available for you to go find with problems. Ideally, from my experience, they'll steer you towards a topic for the PhD, if not outright pick one for you -- otherwise you'll spend at least the first six months with no idea what to do. It takes that long just to absorb enough literature to make a sensible choice. All the PhDs I've seen flounder have been down to the supervisor, usually because he or she is too senior and can't make enough time for the student.
So what's the payoff? For a few years afterwards, I moved into a software engineering role in industry, and was probably paid less overall than if I'd never done the PhD. However with four years under my belt now it has definitely opened doors to me that I wouldn't have been able to get through otherwise, and I'm sure I now out-earn what I could have done, and enjoy more responsibility and control in work. I can say from a career development point of view it's worked out but it's not a night and day difference compared to where I'd be if I'd not done one.
All this, though, is really window dressing. The best bit was, for several years, I walked around knowing more about something that
anyone else in the world. That is utterly intoxicating to me, it's why I did a PhD in the first place, and it's why I would do one again. There is no doubt that it is a stupendous amount of work and stress, and from strictly rational point of view unless you want to work in academia it probably doesn't make sense. But if you're like me, if you'd like to actually create some knowledge, it's a rare opportinity to completely, unequivocally, genuinely be able to say: "There. This is a fact and we know this fact because of me and my work."
The sense of achievement is something I still look back on with great pride, and I'm sure when I look back on my life from my deathbed at some ripe old age [2], the moment my examiner said "I am pleased to say you have passed, Dr. Gaywood" will be one of the highlights.
[1] Those keeping up will note my CV features both a nuclear power firm and an arms firm. This sometimes gets me into trouble at dinner parties when I meet very politically left people.
[2] When I die, I want to go peacefully in my sleep like my grandad -- not kicking and screaming like his passengers.