>noindex in robots meta tags: Supported both in the HTTP response headers and in HTML, the noindex directive is the most effective way to remove URLs from the index when crawling is allowed.

Seems clear enough to me