How to Extract Specific Files and Directories from tar.gz

You do not need to extract an entire .tar.gz archive when you need only one file or directory. List the archive first, then pass the exact member name to tar -xf:

tar -tf archive.tar.gz

Once you have the stored member path, extract only that entry:

tar -xf archive.tar.gz path/to/file.txt

The member name must match the path stored inside the archive, not merely the file's basename. For broader tar options and archive basics, see the tar command cheat sheet.

Quick reference: extract specific content from tar.gz

GNU tar often detects gzip compression when you read a .tar.gz file, so -z is optional on modern systems. This article uses tar -xf and tar -tf consistently; adding -z is equally valid when you want to state the compression format explicitly.

1. List files before extracting

tar works with member names stored in the archive index, not filenames on disk. Start with a plain list:

tar -tf demo-archive.tar.gz

demo-archive/
demo-archive/reports/
demo-archive/reports/report1.txt
demo-archive/reports/report2.txt
demo-archive/reports/report3.txt
demo-archive/file1.txt
demo-archive/file2.txt
demo-archive/link_to_file1.txt

If the listing shows demo-archive/reports/report1.txt, extracting report1.txt alone fails because that basename is not a member name. Use the full stored path instead.

Use verbose listing only when you need permissions, sizes, or symlink targets:

tar -tvf demo-archive.tar.gz

drwxr-xr-x root/root         0 2026-08-15 17:36 demo-archive/
drwxr-xr-x root/root         0 2026-08-15 17:36 demo-archive/reports/
-rw-r--r-- root/root         8 2026-08-15 17:36 demo-archive/reports/report1.txt
-rw-r--r-- root/root         8 2026-08-15 17:36 demo-archive/reports/report2.txt
-rw-r--r-- root/root         8 2026-08-15 17:36 demo-archive/reports/report3.txt
-rw-r--r-- root/root         9 2026-08-15 17:36 demo-archive/file1.txt
-rw-r--r-- root/root         9 2026-08-15 17:36 demo-archive/file2.txt
lrwxrwxrwx root/root         0 2026-08-15 17:36 demo-archive/link_to_file1.txt -> file1.txt

The symlink line shows the stored target (file1.txt). For finding names in a long list, pipe to grep — but use tar -tf output for extraction paths, not parsed tar -tvf columns.

2. Extract one specific file

Pass one member name after the archive filename:

tar -xf demo-archive.tar.gz demo-archive/file2.txt

Add -v when you want tar to print each member as it is extracted:

tar -xvf demo-archive.tar.gz demo-archive/file2.txt

demo-archive/file2.txt

Tar recreates the member's stored directory structure. Extracting demo-archive/file2.txt normally creates ./demo-archive/file2.txt in the current directory, not ./file2.txt. To drop parent directories, use --strip-components in section 8.

3. Extract multiple specific files

List every member you need in one command. Tar extracts only those entries and any parent directories they require:

tar -xvf demo-archive.tar.gz demo-archive/file2.txt demo-archive/reports/report3.txt

demo-archive/file2.txt
demo-archive/reports/report3.txt

You do not need separate tar runs for each file when the member names are already known.

4. Extract a specific directory and its contents

Pass a directory member to pull that tree from the archive:

tar -xvf demo-archive.tar.gz demo-archive/reports/

demo-archive/reports/
demo-archive/reports/report1.txt
demo-archive/reports/report2.txt
demo-archive/reports/report3.txt

Selecting a directory member extracts that directory and the archive members stored beneath it.

5. Extract files with wildcards

GNU tar matches member names with --wildcards. Quote the pattern so the shell does not expand it before tar sees it:

tar -xvf demo-archive.tar.gz --wildcards 'demo-archive/reports/report*.txt'

demo-archive/reports/report1.txt
demo-archive/reports/report2.txt
demo-archive/reports/report3.txt

A leading */ pattern matches the same files when the top-level folder name varies:

tar -xvf demo-archive.tar.gz --wildcards '*/reports/*.txt'

demo-archive/reports/report1.txt
demo-archive/reports/report2.txt
demo-archive/reports/report3.txt

Do not pipe tar -tvf through grep, awk, and xargs for normal wildcard extraction. GNU tar's --wildcards is simpler and avoids breaking on spaces in member names.

6. Extract a symbolic link

Archives can store symlinks separately from their targets. Extract only the link member:

tar -xvf demo-archive.tar.gz demo-archive/link_to_file1.txt

demo-archive/link_to_file1.txt

Tar recreates the symlink with the target path stored in the archive (file1.txt in this demo). If that target file is not present at the expected location, the link is dangling until you restore the target.

Check what tar extracted:

ls -l demo-archive/

lrwxrwxrwx. 1 root root 9 Aug 15 17:36 link_to_file1.txt -> file1.txt

When you need the link to resolve immediately, extract the target file as well:

tar -xvf demo-archive.tar.gz demo-archive/file1.txt demo-archive/link_to_file1.txt

demo-archive/file1.txt
demo-archive/link_to_file1.txt

Confirm both the file and the link resolve in the restored tree:

ls -l demo-archive/

-rw-r--r--. 1 root root 9 Aug 15 17:36 file1.txt
lrwxrwxrwx. 1 root root 9 Aug 15 17:36 link_to_file1.txt -> file1.txt

Tar does not require extracting the target, but a symlink without its target remains broken until the target exists. See create a symbolic link for how links are created on disk.

7. Extract a file into another directory

-C changes where extracted files land. Create the destination first:

mkdir -p /tmp/extracted

Then extract one member into that directory:

tar -xvf demo-archive.tar.gz -C /tmp/extracted demo-archive/reports/report1.txt

demo-archive/reports/report1.txt

Tar still preserves the stored path under the destination, so the file appears at /tmp/extracted/demo-archive/reports/report1.txt. To place only report1.txt directly under /tmp/extracted, combine -C with --strip-components in the next section.

8. Extract without parent directories

--strip-components removes leading path components from each member before writing files. For demo-archive/reports/report1.txt:

• component 1 = demo-archive
• component 2 = reports

Strip two components to leave only report1.txt:

tar -xvf demo-archive.tar.gz --strip-components=2 demo-archive/reports/report1.txt

demo-archive/reports/report1.txt

The verbose line still shows the member name inside the archive; on disk you get report1.txt in the current directory.

Combine with -C to drop parents and choose the output folder:

mkdir -p /tmp/extracted

Run extract with both -C and --strip-components=2 so only the basename lands under /tmp/extracted:

tar -xvf demo-archive.tar.gz -C /tmp/extracted --strip-components=2 demo-archive/reports/report1.txt

demo-archive/reports/report1.txt

The file lands at /tmp/extracted/report1.txt. For more strip examples and zip equivalents, see tar --strip-components.

9. Read a file without extracting it

-O (long form --to-stdout) sends one member to standard output without creating a file:

tar -xOf demo-archive.tar.gz demo-archive/file1.txt

content1

That is useful for quick inspection of configs, logs, manifests, or small scripts. Redirect binary members to a file instead of the terminal:

tar -xOf archive.tar.gz path/file.bin > file.bin

10. Fix Not found in archive

When the member name does not match exactly, GNU tar exits with:

tar: report1.txt: Not found in archive
tar: Exiting with failure status due to previous errors

List members and search for the basename you expected:

tar -tf demo-archive.tar.gz | grep 'report1.txt'

demo-archive/reports/report1.txt

Extract using that full path:

tar -xf demo-archive.tar.gz demo-archive/reports/report1.txt

Some archives store members with a leading ./ (for example ./reports/report1.txt). Use the name exactly as tar -tf prints it, or match with an appropriate --wildcards pattern. When you do not know the full path, --no-anchored lets the pattern match anywhere in the member name:

tar -xvf demo-archive.tar.gz --wildcards --no-anchored '*report1.txt'

demo-archive/reports/report1.txt

Exact member names remain the safest default; pattern flags are for cases where listing first is impractical.

Summary

Selective extraction from .tar.gz follows one workflow: list members with tar -tf, identify the stored path, then pass that path to tar -xf. One command can pull a single file, several named files, an entire directory tree, or wildcard matches when you quote --wildcards patterns for GNU tar.

The usual mistake is passing a basename or a partial path. Tar recreates the directory layout stored in the archive, so a member under demo-archive/reports/ lands under that tree unless you use -C and --strip-components. Use -O to read text members without writing them to disk, and extract symlinks alone when you only need the link entry — they resolve only after their stored target exists on disk.

For full-archive workflows, see ungzip and untar tar.gz. For day-to-day flags and examples, keep the tar command cheat sheet nearby.

References

• GNU tar manual — Extracting Specific Files — exact member-name extraction and -O
• GNU tar manual — wildcard patterns
• GNU tar manual — Modifying File and Member Names — --strip-components and transformed extraction paths