Rel=Canonical to Rewrite or original URL?

[PathInteractive]

Working with a large number of duplicate pages due to different views of products.  Rewriting URLs for the most linked page.  Should rel=canonical point to the rewritten URL or the actual URL? Is there a way to see what the rewritten URL is within the crawl data?

I was taking the approach of rewriting only the base version of each page and then using a rel=canonical on the duplicate pages.  Can anyone recommend a better or cleaner approach?

Haven't seen too many articles on retail SEO when faced with a less than optimized CMS.

Thanks!

[stevenmusumeche]

If you are using rewriting in the web server software (like Apache), the rewritten URL is never exposed to users or search engines.  Please clarify that you are actually rewriting URLs and not just redirecting them to the canonical version.

[PathInteractive]

We are using actual rewrites and then using rel=canonical for all sub pages.