There are far better ways to achieve your goal than using regexen.
Parsing HTML is notoriously fraught with difficulties; the more so, when that HTML is not compliant with well-known standards (4.10 strict; 4.01 loose in particular) That means rolling your own flies in the face of the caution against re-inventing wheels.
To minimize you problems, take a look at the various modules built for the job. A search of CPAN (or ActiveState with ppm if you're on Windows and using AS's Perl) will present a wealth of well-tested and stable (reliable) options.
HTML::Parser, HTML::TableParser, and HTML::Extract are just a few of the many that may suit your needs.