Sorry, I posted my links backwards >.<
The second is much shorter and to the point.
With a sax parser just be aware that it works backwards of the dom (sorta). Instead of loading an XML tree into memory you will trigger events when it comes across the desired matching elements in the tree. This is exceedingly handy because of its low overhead, but as I mentioned it will get complicated if you were to start managing the tree and altering its structure.
Just like everything, both dom and sax have there places. With the size of data you are working with, you won't see a noticeable problem with either, so go with whichever you feel best with using (or are required