>Why count(*) would not be sufficient? If you have lots of fields, the result would be more than 255 chars, so it would not work.

DOH! That was dumb, the SELECT COUNT has no bearing on the actual count. It is the GROUP BY that directly affects your count, and, therefore, data types are irrelevant. I see no better or more effecient way to find duplicates by multiple field values than what you came up with.