It depends on what you're trying to do and how newless of a clubie that you are.  If you're not up to speed on DOM's (or SAX for that matter) then do your research on that first.  If you're up to speed on DOM then get the JavaDocs for Xerces and look over them for a few minutes.  You'll see quickly that they parallel the DOM exactly.  If you need more insight, there's some good articles at http://www.javaworld.com (use the search).

If you're XML savvy, you might want to consider using XPath and DOM4J.  XPath takes all of the tediousness out of walking the DOM if you're just looking for certain nodes.  DOM4J is just MUCH easier to use with XPath (and, well, DOM for that matter).

Well I've found the built-in DOM APIs in Java 1.4 to be very easy to use and straightforward.  I wrote a couple helper methods that build on it to simplify some things.. but that's about it.

JAXB - Java Architecture for XML Binding existed to do exactly this.. pass the XML scheme through a processor and out popped a bunch of parsing objects.

http://java.sun.com/xml/jaxb/

Kev

Your focus on Xerces in particular is misguided. Xerces is just one parser that implements JAXP. What you should be focusing on is learning JAXP. It's like saying you're going to "learn how to operate a toyota" instead of "learn to drive a car".

Did you look at Sun's tutorial on JAXP? It's pretty good.

JDOM is just fine, it's a nice library. It's not as popular because it's proprietary, is not backed by Sun, and is Java only.

No, you are thinking of DOM vs. SAX.  JAXP has implementations for both DOM and SAX I believe.
DOM does represent the document as a tree of nodes in memory.  SAX is event based and works on the fly as it reads the XML data from the source.