To identify existing pages and to avoid invalid pages

I have to identify whether a particular url is existing or not. For that, using curl i can retrieve the contents of the page. If a particular domain is not an existing one, then I can obtain the error msg "Couldn't resolve host 'www.dasdas.com' " and corresponding error code using curl, so that i can trace it properly. But if a domain exist, and there will be a common error page for showing pages which are not found, the curl will give the contents in the corresponding error page, and will treat it as an existing page. There will not be any error message or error code in this case. So how can I identify properly whether the particular link inside an existing domain is an existing link or not....