You might be better off using HTML::Parser to parse out HTML tags, then apply your regular expression on a tag-by-tag basis. It will be very difficult to get your regular expression to work properly considering the variety and complexity that can occur in an HTML document.