I understand your concern of parsing html with regexes, but this string will never be a full set of html. It will have some html (of course with the para's) but using a parser would be overkill for this situation.