Is it worth catering for version 3 browser users?
Note: This is Jukka Korpela's typically well thought-out and informative reply (with incidental humour), copied in full.
It depends on what you mean by that. I'm having hard time trying to imagine what measures I need to take to _exclude_ them. Why do you think "v3" would be a problem in authoring? Most of the newer HTML constructs (those not supported by NS 3 or IE 3) have been designed with graceful fallback in mind, and all an author needs to do is to specify the fallback alternative, which should be done anyway, for the benefit of other user agents, current and future.
To take a simple example, consider IFRAME. It's supported by IE 3 but not by NS 3; neither is it supported by NS 4, or by many other browsers. It's hard to tell what indexing robots do with it, but I'm pretty sure that not all of them check what's behind the SRC attribute, so if you include something via IFRAME SRC only, several search engines might never find it. Moreover, some advanced browsers already let the user _disable_ IFRAMEs, i.e. they support IFRAME but they are also capable of using the fallback instead (and there are different reasons why this is a good option).
And by providing the fallback content, which in this case means writing some useful content between <iframe ...> and </iframe> content, you accommodate quite a many browsing and other situations, not just NS 3. The fallback content could be as simple as a normal link to the document you would like to have IFRAMEd, or a copy of its content, or something else.
Is there a reliable source that indicates how many people still use version 3 browsers?
Reliable estimates cost real money. Bogus data is available for free, especially on the Internet. It might be pretty safe that the number of NS and IE v3 users is below some small percentage, say < 5 % on the Web as a whole, though this is still just an educated guess. And it is a relatively unimportant guess; there is hardly any reason to make it.
It might make sense to ask: How many people still use IE 3 _with style sheets enabled_? The reason for asking this would be the fact that CSS "support" in IE 3 is so poor that if you don't use features that make CSS-enabled IE 3 users suffer, you can hardly do anything useful in CSS. But if you don't take the position that the problem really needs to be fixed where it is (i.e., if you use IE 3, disable style sheets) and people assumably have got the lesson in this case by now, then the sensible approach is to turn the question to a technical CSS question: How could I hide my style sheet from IE 3?
(Actually, Netscape 4 with style sheets enabled is a more important practical problem at present.)