We have one rather large content database in our environment. In this content database, a single site collection resides. In Central Admin, this site collection lists as being 464699 MB big. Since this is is the only site in the database, I'd expect the database to be roughly equal in size. It is not however, actually the SQL MDF file is about 800GB's. What could be causing the extra overhead and how do I limit it? I checked with our DBA's, there is no air / shrinking won't help.
I know we're not exactly following best practices and we shouldn't go over 200GB's, but that's a different story.
Update: i also ran Get-SPSiteAdministration to check the DiskUsed property of the site, that returned 487317920421.
Update: the recycle bin data as mentioned in some answers is not the culprit. The second stage bin is nearly empty and first stage is also not taking up much space.
Update: our DBA ran the report to show the table sizes. The result is pretty weird. The AllDocStreams table is 535004448 KB (= 522465MB) big. The AllDocVersions table: 270222MB. Central Admin is still giving me 479219MB for the entire site collection, which is less than both numbers individually. Audit table is only 15418MB, so that's not the major culprit.
Conclusion: most space is used by documents, not even the olders versions. And that space alone is more than what Central Admin is telling me. As already stated, the recyclebins are pretty much empty so those cannot be the issue either. So now the question is: is the number in Central Admin just wrongly calculated, or is there some left over / orphaned data in those tables which is not calculated but still there for some reason? And if that would be the case: how can I get rid of it?