While this SAX event based parser is better for memory management than the tree based parsers of SimpleXML and DOM, the pull-based parser XMLReader is much easier to use than the xml_parser_* functions, and still doesn't require loading the file into memory.
An event-based parser such XML Parser is preferable for large files, because tree-based parsers must fully load the file into memory in order to parse the XML. Event-based parsers do not need to load the entire file into memory to begin parsing.
If you're wanting to actually work with XML data as it was intended, treating it as a tree, try http://us2.php.net/manual/en/book.simplexml.php .