ntfsundelete - recover a deleted file from an NTFS volume.
ntfsundelete [options] device
ntfsundelete has three modes of operation: scan, undelete and copy.
The default mode, scan simply reads an NTFS Volume and looks for files that have been
deleted. Then it will print a list giving the inode number, name and size.
The undelete mode takes the files either matching the regular expression (option -m) or
specified by the inode-expressions and recovers as much of the data as possible. It
saves the result to another location. Partly for safety, but mostly because NTFS write
support isn't finished.
This is a wizard's option. It will save a portion of the MFT to a file. This probably
only be useful when debugging ntfsundelete
ntfsundelete only ever reads from the NTFS Volume. ntfsundelete will never change the
ntfsundelete cannot perform the impossible.
When a file is deleted the MFT Record is marked as not in use and the bitmap representing
the disk usage is updated. If the power isn't turned off immediately, the free space,
where the file used to live, may become overwritten. Worse, the MFT Record may be reused
for another file. If this happens it is impossible to tell where the file was on disk.
Even if all the clusters of a file are not in use, there is no guarantee that they haven't
been overwritten by some short-lived file.
In NTFS all the filenames are stored as Unicode. They will be converted into the current
locale for display by ntfsundelete. The utility has successfully displayed some Chinese
pictogram filenames and then correctly recovered them.
Extended MFT Records
In rare circumstances, a single MFT Record will not be large enough to hold the metadata
describing a file (a file would have to be in hundreds of fragments for this to happen).
In these cases one MFT record may hold the filename, but another will hold the information
about the data. ntfsundelete will not try and piece together such records. It will sim-
ply show unnamed files with data.
Compressed and Encrypted Files
ntfsundelete cannot recover compressed or encrypted files. When scanning for them, it
will display as being 0% recoverable.
The Recovered File's Size and Date
To recover a file ntfsundelete has to read the file's metadata. Unfortunately, this isn't
always intact. When a file is deleted, the metadata can be left in an inconsistent state.
e.g. the file size may be zero; the dates of the file may be set to the time it was
deleted, or random.
To be safe ntfsundelete will pick the largest file size it finds and write that to disk.
It will also try and set the file's date to the last modified date. This date may be the
correct last modified date, or something unexpected.
Below is a summary of all the options that ntfsundelete accepts. Nearly all options have
two equivalent names. The short name is preceded by - and the long name is preceded by
--. Any single letter options, that don't take an argument, can be combined into a single
command, e.g. -fv is equivalent to -f -v. Long named options can be abbreviated to any
unique prefix of their name.
-b, --byte NUM
If any clusters of the file cannot be recovered, the missing parts will be filled
with this byte. The default is zeros.
When scanning an NTFS volume, any filename matching (using the --match option) is
case-insensitive. This option makes the matching case-sensitive.
-c, --copy RANGE
This wizard's option will write a block of MFT FILE records to a file. The default
file is mft which will be created in the current directory. This option can be
combined with the --output and --destination options.
-d, --destination DIR
This option controls where to put the output file of the --undelete and --copy
This will override some sensible defaults, such as not overwriting an existing
file. Use this option with caution.
Show a list of options with a brief description of each one.
-i, --inodes RANGE
Recover the files with these inode numbers. RANGE can be a single inode number,
several numbers separated by commas "," or a range separated by a dash "-".
-m, --match PATTERN
Filter the output by only looking for matching filenames. The pattern can include
the wildcards '?', match exactly one character or '*', match zero or more charac-
ters. By default the matching is case-insensitive. To make the search case sensi-
tive, use the --case option.
Recover parts of the file even if they are currently marked as in use.
-o, --output FILE
Use this option to set name of output file that --undelete or --copy will create.
Display the parent directory of a deleted file.
-p, --percentage NUM
Filter the output of the --scan option, by only matching files with a certain
amount of recoverable content. Please read the caveats section for more details.
Reduce the amount of output to a minimum. Naturally, it doesn't make sense to com-
bine this option with --scan.
Search through an NTFS volume and print a list of files that could be recovered.
This is the default action of ntfsundelete. This list can be filtered by filename,
size, percentage recoverable or last modification time, using the --match, --size,
--percent and --time options, respectively.
The output of scan will be:
Inode Flags %age Date Size Filename
6038 FN.. 93% 2002-07-17 26629 thesis.doc
|Flag Description |
|F/D File/Directory |
|N/R (Non-)Resident data stream |
|C/E Compressed/Encrypted data stream |
|! Missing attributes |
The percentage field shows how much of the file can potentially be recovered.
-S, --size RANGE
Filter the output of the --scan option, by looking for a particular range of file
sizes. The range may be specified as two numbers separated by a '-'. The sizes
may be abbreviated using the suffixes k, m, g, t, for kilobytes, megabytes, giga-
bytes and terabytes respectively.
-t, --time SINCE
Filter the output of the --scan option. Only match files that have been altered
since this time. The time must be given as number using a suffix of d, w, m, y for
days, weeks, months or years ago.
If ntfsundelete is confident about the size of a deleted file, then it will restore
the file to exactly that size. The default behaviour is to round up the size to
the nearest cluster (which will be a multiple of 512 bytes).
Select undelete mode. You can specify the files to be recovered using by using
--match or --inodes options. This option can be combined with --output, --destina-
tion, and --byte.
When the file is recovered it will be given its original name, unless the --output
option is used.
Increase the amount of output that ntfsundelete prints.
Show the version number, copyright and license for ntfsundelete.
Look for deleted files on /dev/hda1.
Look for deleted documents on /dev/hda1.
ntfsundelete /dev/hda1 -s -m '*.doc'
Look for deleted files between 5000 and 6000000 bytes, with at least 90% of the data
recoverable, on /dev/hda1.
ntfsundelete /dev/hda1 -S 5k-6m -p 90
Look for deleted files altered in the last two days
ntfsundelete /dev/hda1 -t 2d
Undelete inodes 2, 5 and 100 to 131 of device /dev/sda1
ntfsundelete /dev/sda1 -u -i 2,5,100-131
Undelete inode number 3689, call the file 'work.doc' and put it in the user's home direc-
ntfsundelete /dev/hda1 -u -i 3689 -o work.doc -d ~
Save MFT Records 3689 to 3690 to a file 'debug'
ntfsundelete /dev/hda1 -c 3689-3690 -o debug
There are some small limitations to ntfsundelete, but currently no known bugs. If you
find a bug please send an email describing the problem to the development team:
ntfsundelete was written by Richard Russon and Holger Ohmacht, with contributions from An-
ton Altaparmakov. It was ported to ntfs-3g by Erik Larsson and Jean-Pierre Andre.
ntfsundelete is part of the ntfs-3g package and is available from:
ntfs-3g 2011.4.12AR.4 November 2005 NTFSUNDELETE(8)