For big files DOM is out of the question though there always tricks of course. With SAX I've parsed big files with good performance (Personally I favor the Xalan and Xerces implementations of Apache). Although I do use libxml2 I invariably use XML::Twig when I am in a Perl environment. I have parsed files over 1 GB with it. There is also XML::Twig::XPath but I never used it. You would have to check if it solves your problem. The document I mentioned is only one from a (big) series. They elaborate the best practices for XML-Schema usage.