Ed Loehr <pggeneral@bluepolka.net> writes:
> Does pg_class.reltuples include "dead" tuples awaiting vacuuming?
No, at least not when it's been updated by VACUUM; since the VACUUM
will have removed said tuples.
I think it probably does count zombie tuples (ones that are committed
dead, but are still visible to some open transaction so VACUUM couldn't
remove 'em).
regards, tom lane