Really you shouldn’t need it. What you should be doing is avoiding duplicate content altogether. Every piece of content on your site should exist on one permanent, unique URL, and any duplicate pages should be consolidated through 301 redirects. But for a variety of business, technical or editorial reasons, sometimes publishing the same content or resolving the same page on more than one URL simply can’t be avoided.
On newspaper and magazine sites the main culprit is typically the practice of appending tracking codes to the end of URLs, for example: www.yoursite.com/article?xid=rss (or ?xid=topstories or ?cid=partner, etc…).
Fortunately the major engines came to the rescue in February with rel=canonical which allows sites to identify which URL is the canonical or primary version of a page. It is placed in the <head> of all versions of a particular page and points to the canonical URL for the page:
<link rel="canonical" href="http://www.example.com/canonical-URL"/>
Google describes rel=canonical as a “hint that we honor strongly” as opposed to a firm directive, so the use of the tag does not guarantee that all duplicate content issues will be resolved.
However it greatly increases the likelihood that the canonical version of a page will be the one displayed in the search results, and if it functions as described, it will also transfer the value of links pointing to the duplicate pages to the canonical URL.
This means that publishers can append tracking codes to URLs as needed while avoiding the duplicate content issues that place the content at a competitive disadvantage.
It is still early days for the tag but based on results from some magazine sites we work with that have been experimenting with it, I’d say so far, so good.