On the assumption that you will be doing this for many links in (possibly) many files, I strongly suggest abandoning the regexp in favor of a module (something in the HTML::Parser family). Those modules have tested for circumstances that might be or have been overlooked (for example, the a\s or a\b problem tackled by sauoq and the pusher robot. I'm sure you'd rather strip those tags instead of revising a continuously growing regexp everytime you hit a bump in the road.

John J Reiser
newrisedesigns.com