You're most likely getting bitten by greediness and/or the fact that trying to parse HTML with just regexen is bound to end in pain and tears. Use HTML::TreeBuilder or HTML::TokeParser (or one of the derivatives) instead.