There have been some experiments performed for SEO purposes which indicate that at least the big players (Google, for example) can and do follow some simple JavaScript. They avoid sneaky redirects and such, but some basic content manipulation does seem to get through. (I don’t have a link handy for Google themselves confirming or denying this, it’s just various posts I’ve come across when dealing with this before.)
However, this is generally considered unreliable. If SEO is being done for any important purpose, don’t rely on the spiders indexing much dynamic content.
There’s actually a very good (in my opinion, anyway) answer here to a very similar question. What I like about that answer is how it breaks down the steps for generating good, indexable, and best of all maintainable web pages with concerns properly separated. Adhering as much as possible to this process generally results in good SEO, good accessibility, and good design skills in general.