Make your 404 page a php script, for example error.php, and include something like this inside the body:

<?php
 $url = $_SERVER[ "REQUEST_URI" ];
 $referrer = $_SERVER[ "HTTP_REFERER" ];
 if ( $referrer == "" )
 $referrer = "Unknown";
 if ($url!= "/sitemap.rdf" 
&& stristr($url, '/_vti_bin/') == FALSE
&& stristr($url, '/siteinfo.xml') == FALSE
&& stristr($url, '/MSOffice/clt') == FALSE
){
 mail("badpage@youremail.com", "Page Not Found", 
 "Requested Page: " . $url 
 . "\r\nReferred By: " . $referrer
 . "\r\nRemote Addr: " . $_SERVER["REMOTE_ADDR"] . " (" . $_SERVER["REMOTE_HOST"] . ")"
 . "\r\n"
 . "Cookies: \r\n"
 . implode(",", $_COOKIE)
 . "\r\nRequest URI: " . $_SERVER["REQUEST_URI"]
 . "\r\n"
,"From: badpage@youremail.com" );
 }
?>

that will send you an email for the problem pages.
obviously you also need to output some html.

Gert

Off-topic, but urgent:

Your ErrorDocument code is ill-formed, and will result in a 200-OK response for all Not Found and Forbidden resources. This can lead to serious problems with search ranking and results.

The proper form for ErrorDocument specifies a local URL_path, not a canonical URL. If a canonical URL is specified, Apache will produce a 302-Found redirect response, and the client will then receive a 200-OK when it re-requests the document.

You can easily confirm this behaviour using the "Live HTTP Headers" extension for FireFox/Mozilla, or by using an online server headers checker.

Your code should read:

ErrorDocument 404 [b]/error.htm[/b]
ErrorDocument 403 [b]/error.htm[/b]

For more information, read the detailed notes in the documentation for the ErrorDocument [httpd.apache.org] directive.

Jim

Thanks for the replies.

Gert:
I won't need an email, but I'm looking at your code.
Can you say what the sitemap.rdf, _vti_bin (MS Frontpage?), siteinfo.xml and the MSOffice statements do?

Jim:
I had tried a relative address when I started setting up 404 pages for my sites some years ago. But then the only thing that seemed to work was using an absolute address. I've just tried /error.htm and it now seems to work?!

NB. I appreciate your post, but maybe the SE's are cleverer than 'we' think as I use this syntax on over 80 domains and they all do very well on Google etc.

Thanks again - I'm still tinkering with both of your suggestions.

Having an email sent for each 404 isn't really what I was after.

I've just realised I would have been better posting this in the "Tracking and Logging" forum.

Is it possible for a mod to move it?