I'd use XML::LibXML, and select the nodes you want using XPath.

#!/usr/bin/perl

use strict;
use warnings;

use XML::LibXML;

my $parser = XML::LibXML->new;

my $dom = $parser->parse_file( 'test.xml' ); # load file

my @wanted_nodes = $dom->findnodes( './/an/xpath/here' ); # select

print $_->toString for @wanted_nodes; # output
[download]

merlyn has written a column on using XML::LibXML with HTML, which may be of interest.