404's in WMT are old pages and referrer links no longer linking to them.

[rock22]

Within the last 6 days, Google Webmaster Tools has shown a jump in 404's - around 7000.  The 404 pages are from our old browse from an old platform, we no longer use them or link to them.

I don't know how Google is finding these pages, when I check the referrer links, they are either 404's themselves or the page exists but the link to the 404 in question is not on the page or in the source code.  The sitemap is also often referenced as a referrer but these links are definitely not in our sitemap and haven't been for some time.  So it looks to me like the referrer data is outdated.  Is that possible?

But somehow these pages are still being found, any ideas on how I can diagnose the problem and find out how google is finding them?

[MikeRoberts]

Having a similar problem with a new site that was created by copying an old site in its entirety. Went through the trouble of cleaning everything up, having pages that were no longer relevant removed, fixed the sitemaps, etc. and now months later WMT showed me a spike of 404s for the old pages with the referrers as the XML sitemap and sitemap page... but they are definitely not be linked from there. I'm assuming there was some sort of hiccup with Google using an older, cached version of the sitemap to find these links.

I wound up just clearing the errors out of WMT and waiting to see if it will recrawl the error pages again. If Google continues to crawl them even though they aren't being linked to, then our next course of action was going to be 301ing them all just in case.

[sprynewmedia]

How long ago did you switch platforms?  It can take months for Google to come back around to a page that linked to your site.  Page on your site will stay in the cache until a few passes.

When you switch, did you do any 301 redirects?  Examine the back links to your domain - any that come from good pages should be redirected to the new URL.  If not, they will be scooped up by active SEOs. (finding 404 links is a popular link building technique).

http://support.google.com/webmasters/bin/answer.py?hl=en&answer=93633

If you know the links will be dead forever, try using a 410 response as it is supposed to make search engines drop the page faster.

http://www.seroundtable.com/404-410-google-15225.html (bottom)

Have you requested Google remove old directories/pages?  If the content is gone and has no back links, try a removal request.

http://support.google.com/webmasters/bin/answer.py?hl=en&answer=1663427