What .tar and .tar.gz actually are
A .tar file is not compressed. It is a tape archive: a flat run of 512-byte blocks, one header block per member followed by that member’s data padded to the next block boundary. The header is fixed-offset ASCII — name at byte 0, size at byte 124 written in octal, type at byte 156 — and there is no index anywhere, which is why reading a tar means walking it from the beginning.
Because tar does not compress, it is almost always wrapped: .tar.gz (also written
.tgz) is a tar file compressed with gzip, and the two steps are separate. That is the whole
reason the format confuses people on Windows — unpacking it is genuinely two operations, and the
built-in tools historically did neither.
How to extract a tar.gz file online
- Open the archive. Drop the .tar.gz, .tgz or .tar onto the box above, or click to browse. Gzip is unwrapped automatically, so there is no separate decompress step.
- Look at what is inside. Every member is listed with its type, uncompressed size, modification date and Unix permissions. Folders, symlinks and hard links are labelled rather than hidden.
- Save the files you want. Save writes the original bytes of that member. Files are extracted individually, so a 300-file archive does not become 300 downloads unless you ask for them.
- Keep the listing. Copy file list puts the whole contents on your clipboard; Save list as CSV writes name, type, size, mode and date to a spreadsheet-ready file.
What comes out
- Every member of the archive — files, folders, symlinks, hard links and device nodes, each labelled with what it is.
- True uncompressed sizes, read from the header rather than estimated.
- Unix permissions in octal, and the owner name where the archive recorded one.
- Modification dates, converted from the Unix timestamp in the header.
- Symlink targets, so you can see what a link pointed at before extracting.
- The compression ratio, once the gzip layer is off.
Long paths are read correctly. Both extensions that carry them are handled — GNU long names,
where the real path is stored as its own pseudo-member, and PAX extended headers, which store
path and size as key-value records. Without those, any path over 100
characters comes out truncated.
What is supported, and what is not
Supported: plain .tar, and .tar.gz, .tgz and
.gz. Gzip is decompressed by the browser’s own DecompressionStream, which
means there is no library to download — the code that unpacks your archive is the same code the
browser uses for every compressed web page it loads.
Not supported: .tar.xz, .tar.bz2 and
.tar.zst. XZ uses LZMA2, bzip2 uses Burrows–Wheeler, Zstandard uses its own scheme,
and no browser implements any of them. Each is detected by its signature and named, with instructions,
rather than failing with a meaningless error. Decompress one of those to a plain .tar first
and it will open here.
Sizes above 8 GB, which GNU writes in a binary form rather than octal, are read correctly — though the 512 MB working limit below matters long before that does.
Why the archive is never uploaded
The file is read from your disk, decompressed by the browser and walked in memory. No server takes part, so the archive never leaves your device.
This is the practical difference from the usual online extractor. Tar archives are how server backups, database dumps, configuration directories and source trees get moved around — the contents are frequently things that must not be handed to a stranger’s server. Here there is no upload, so there is no decision to make about whether the site can be trusted with them.
What this tool does not do
Four boundaries:
- It does not create or modify archives. This reads, it does not write tar.
- It does not extract everything as one folder. Browsers cannot write a directory tree to disk, so files are saved individually.
- It does not handle xz, bzip2 or zstd. Those need decoders far larger than the tool itself.
- It does not restore permissions or ownership. Those are shown, because they are often what you opened the archive to check, but a downloaded file gets your own system’s defaults.
Archives over 512 MB are refused rather than crashing the tab. A truncated archive — usually an interrupted download — is read as far as it goes and clearly marked as incomplete, instead of silently listing fewer files than it should.
Why people need to open a tar.gz
- On Windows with nothing installed — a release downloaded from GitHub or a Linux project arrives as .tar.gz and nothing on the machine will open it.
- Checking a backup — confirming a server backup actually contains what it should, before trusting it.
- Taking one file out of a source release without unpacking the whole tree.
- Reading permissions — seeing what mode and owner a deployment archive will restore.
- On a locked-down machine where installing 7-Zip is not an option.
tar.gz compared with ZIP, RAR and 7z
The key difference is architectural, not cosmetic. A ZIP compresses each file separately and keeps a central directory at the end, so any single file can be pulled out without touching the rest. A .tar.gz compresses the whole archive as one stream, which gives better compression across similar files but means the entire thing must be decompressed to reach anything inside it.
For .zip archives, use the ZIP extractor. RAR and 7z are not supported anywhere on this site and not by any browser-based tool that is honest about it: both need large proprietary or LZMA decoders, so use 7-Zip or WinRAR on the desktop.
Once a file is out, the CSV column extractor, the JSON field extractor and the Excel data extractor read what is inside it.
For a full format comparison and decision guide, read ZIP vs TAR vs GZ vs 7Z.
TAR format standards and extraction edge cases
TAR is defined in POSIX.1-2017 (pax extended header format). Older archives use the ustar format (POSIX.1-1988) or the original V7 format. Each file entry is preceded by a 512-byte header block: filename (100 bytes, NUL-padded), file mode, UID/GID, file size as octal ASCII, mtime as Unix seconds in octal, a checksum, and a type flag byte.
Three edge cases: (a) filenames longer than 100 bytes use the POSIX ustar prefix field
(155 bytes) or a GNU longname extension — a type-flag L ("L") header
immediately before the entry carries the full path; both methods are resolved to the
complete filename; (b) a TAR.GZ stream must be fully decompressed before the entry list is
known — unlike a ZIP's end-of-central-directory, there is no up-front index, so a very large
TAR.GZ may be slow to open; (c) hard links and symbolic links are type-flag entries 1 and 2
respectively, with the link target stored in the header — they are listed with their target
path, but there is no file data in the archive to save. For a full comparison of when to use
each format, see ZIP vs TAR vs GZ vs 7Z.
Frequently asked questions
How do I extract a tar.gz file without WinRAR or 7-Zip?
Drop it onto this page. The gzip layer is removed and the archive walked in your browser, so nothing needs installing and nothing is uploaded.
What is the difference between .tar, .tar.gz and .tgz?
A .tar is an uncompressed bundle of files. A .tar.gz is that bundle compressed with gzip. .tgz is simply a shorter spelling of .tar.gz, for systems that dislike two extensions. All three open here.
Can I extract tar.xz or tar.bz2?
No. XZ uses LZMA2 and bzip2 uses Burrows-Wheeler, and no browser can decompress either. The tool recognises them and says so. Decompress to a plain .tar first and it will open here.
Is my archive uploaded anywhere?
No. It is decompressed and read inside your browser. This matters for tar in particular, since server backups and config directories usually travel this way.
Can I extract everything at once into a folder?
No. Browsers cannot write a folder structure to disk, so files are saved one at a time. The full listing can be exported as CSV if you need a record of everything.
Why does Windows not open .tar.gz on its own?
Modern Windows 11 can, via tar on the command line, but File Explorer still has no menu item for it, and older versions have no support at all. It is a Unix format that never got a graphical home on Windows.
Does it show file permissions?
Yes, in octal, along with the owner name where the archive stored one. That is often exactly what you are checking before deploying an archive.
What if my file is only .gz and not .tar.gz?
A plain .gz holding a single non-tar file will decompress but then fail the tar checksum, and the tool says so rather than showing nonsense.
Why does it say the archive is truncated?
Because the file ends earlier than its own headers describe, which almost always means the download stopped early. Everything readable up to that point is still listed.