Also consider XML::LibXML, which despite its name, can be coaxed into reading HTML, and then provides DOM and XPath interfaces into your HTML tree. It's also far faster than HTML::Tree, keeping the tree in C space, only converting to Perl scalars when necessary.

I wrote a column about using it to extract data from a web page.

-- Randal L. Schwartz, Perl hacker
Be sure to read my standard disclaimer if this is a reply.