why is the df command failing me?

See the following log:

root@company-mirror:/mirror/hin-file# df -m
Filesystem           1M-blocks      Used Available Use% Mounted on
/dev/sda1               935843    893976         0 100% /
varrun                     498         1       498   1% /var/run
varlock                    498         1       498   1% /var/lock
udev                       498         1       498   1% /dev
devshm                     498         0       498   0% /dev/shm
root@company-mirror:/mirror/hin-file# du -ch 2008-02-01/ | grep total
60K    2008-02-01/admin/Desktop Backup/Old Desktop Files/total results
20G    total
root@company-mirror:/mirror/hin-file# rm -rf 2008-02-01
root@company-mirror:/mirror/hin-file# df -m
Filesystem           1M-blocks      Used Available Use% Mounted on
/dev/sda1               935843    893889         0 100% /
varrun                     498         1       498   1% /var/run
varlock                    498         1       498   1% /var/lock
udev                       498         1       498   1% /dev
devshm                     498         0       498   0% /dev/shm
root@company-mirror:/mirror/hin-file#

I have been doing this all day.  Deleting data yet it still says the hard drive is at max capacity(100%)

I need to get this figured out.....

UPDATE 12/30/2008:  The directory/files in question are the ones affect by our backup script.  We use rsync and cp -al(l being the important option here) to backup up our servers incrementally every night.  I did not design the script, but it is slick and explains the previous output.  I was not deleting a 20G directory, I was deleting a directory that pointed to the same inodes that comprise the 20G.  Or, in other words, they were hard links.

Now I just need to figure out how to see all the hard links in a directory....

UPDATE 12/31/08: After digging more deeply into hard links, it was pointed out to me that every file is a hard link.  SO I had to readjust my approach to take that into consideration.   Since I had a directory that had folders matching up to the daily backups, I found a unique file that hasn't changed in all those days(and ancient IT doc), and did a find . -samefile /file/path/filename  I then matched that against the existing directory content which clued me into which directories where entire unique directory, rather than cp -al increments.  If that makes any sense, fine, if not and if this can help you or you're just curious email me :p