I think it's a bit better than, say, physics. The market seems like it has been going downhill, though. It seems from the recent AMS data I've seen, that about 2/3s of math PhDs get SOME sort of academic jobs (the other 1/3 in industry and some unemployed or under-employed, such as yours truly at the moment). What's not clear from that data, though how many of these new PhDs are have the terrible adjunct positions, how many are postdocs and will those people find a permanent place?
Here's a relevant post.
http://mathoverflow.net/questions/23716/how-hard-is-it-to-get-tenure-in-mathematics
Personally, I was screwed because I sucked at teaching and the department didn't let me teach my own class (except once), so I never had the chance to address those issues sufficient. Plus, I completely lost interest in the very idea of doing research on "cutting-edge" topics once I saw what it was like. I had always assumed that I could just keep learning and think about math and things would work out, since I was apparently pretty good at it, but they didn't even come close to working out. I found that my interests were completely orthogonal to anything that would give me much credit in academia, at least within the foreseeable future.
I suppose the moral of the story is to try to really think about what it would be like to be a math professor and if you'd actually like it. This can be hard to do. As far as teaching goes, tutoring and practicing public speaking (toastmasters, etc.) is a really good start. I didn't tutor until after I taught my first class, and that would have helped a lot. For research, there are always REUs and that sort of thing, but you can always do your own project. Write some long expository article about an interesting topic in LaTeX, try to come up with your own ideas and fool around with them. That sort of thing. But what that doesn't tell you about is the nature of what mathematicians are actually studying, which takes a very long time to understand. As it happens, it was perhaps this absurd level of complexity that I found more off-putting than anything (specifically, there was just too much material to understand it all as deeply as I'd like, and it's not just me--lots of mathematicians have to take the more complicated stuff on faith because it's just to time-consuming to go through it all). To my mind, it wasn't a good kind of "deep" complexity. More like a horrendous sort of "fall down the bottomless pit of advanced mathematics" sort of complexity. I don't know. Maybe I shouldn't have chosen topology.
Another moral of the story is to come up with a back-up plan.
Saying you like math enough to get a math PhD is a bit like saying you like pizza so much that you would like it if some locked you up in a room and wouldn't let you out until you ate 20 extra-large pizzas in one sitting.
