The cross-platform archive format everyone's computer already understands Created by Phil Katz (PKWARE) in 1989 as an open alternative to ARC. The only common archive format natively readable on Windows, macOS, and Linux without extra software.
- What are zip and unzip?
- Where do they live?
- How the ZIP format works internally
- zip — Syntax and Usage
- unzip — Syntax and Usage
- Compression Levels
- Password Protection & Encryption
- All Key zip Options
- All Key unzip Options
- zip vs tar.gz vs 7z
- Working with Large Files (Zip64)
- Related Commands
zip creates compressed archive files in the ZIP format, and unzip extracts them. Unlike tar, which is purely an archiver that's typically paired separately with a compressor (like gzip), ZIP is a single format that does both archiving (bundling multiple files/directories into one) and compression (shrinking file size) at once, per-file rather than for the whole archive as a single stream.
zip archive.zip file1.txt file2.txt # create
zip -r archive.zip myfolder/ # create, recursively
unzip archive.zip # extract
unzip -l archive.zip # list contents without extractingWhy ZIP is still everywhere in 2026: it's the only archive format with truly universal, built-in OS support — Windows Explorer, macOS Finder, and virtually every Linux file manager can open a .zip with zero additional software. This makes it the default choice whenever the recipient's environment is unknown or non-technical (email attachments, downloadable assets, cross-platform data exchange).
/usr/bin/zip
/usr/bin/unzip
which zip unzip
zip --version
unzip -vOn many minimal Linux installations, zip/unzip are not installed by default (unlike tar, which almost always is) and must be installed separately:
sudo apt install zip unzip # Debian/Ubuntu
sudo dnf install zip unzip # Fedora/RHEL
sudo pacman -S zip unzip # Arch
brew install zip unzip # macOS (usually already present)Both are part of the Info-ZIP project, the de facto standard open-source implementation.
A ZIP archive stores each file independently compressed, with its own local file header, compressed data, and CRC-32 checksum:
[Local File Header][Compressed Data][Data Descriptor] ← file 1
[Local File Header][Compressed Data][Data Descriptor] ← file 2
...
[Central Directory] ← index of all entries
[End of Central Directory Record] ← points to Central Directory
This has two major practical consequences compared to tar.gz:
- Random access — you can extract, list, or verify a single file from a huge ZIP archive without reading through the entire archive first, because the Central Directory at the end tells
unzipexactly where each file's data starts. - Slightly worse compression ratio — because each file is compressed in isolation, ZIP can't exploit redundancy between files the way
tar.gzcan (which compresses the whole concatenated stream together). Many small, similar files compress noticeably better astar.gzthan as.zip.
Located at the end of the file, the Central Directory is why tools like unzip -l can instantly list contents of even a multi-gigabyte archive — they seek to the end, read the Central Directory, and never need to scan the compressed data itself just to see filenames.
# This is instant even on a huge archive, because it only reads the
# Central Directory at the end of the file:
unzip -l huge_archive.zipThis also means a ZIP file with a corrupted or missing Central Directory (e.g., a download that was cut off) can be much harder to recover from than a similarly-truncated tar.gz, since standard tools expect to find that index at the end.
Each file entry records its own compression method — most commonly Deflate (method 8), but entries can also be stored uncompressed (method 0, "stored"), which is used automatically for files that don't compress well (already-compressed formats like .jpg, .mp4, .zip itself) to avoid wasting CPU for no size benefit.
zip [OPTIONS] ARCHIVE.zip FILE...# Create a new archive from specific files
zip archive.zip file1.txt file2.txt
# Add a directory and everything inside it (recursive)
zip -r archive.zip myfolder/
# Add files to an EXISTING archive
zip archive.zip newfile.txt
# Update only files that are newer than what's already in the archive
zip -u archive.zip file1.txt
# Delete a specific file from an existing archive
zip -d archive.zip unwanted_file.txt
# Exclude specific patterns while creating
zip -r archive.zip myfolder/ -x "*.log" -x "*/.git/*"
# Move files INTO the archive, deleting the originals afterward
zip -rm archive.zip myfolder/unzip [OPTIONS] ARCHIVE.zip [FILE...] [-d DESTINATION]# Extract everything into the current directory
unzip archive.zip
# Extract into a specific destination directory
unzip archive.zip -d /path/to/destination
# List contents WITHOUT extracting
unzip -l archive.zip
# Extract only specific files/patterns
unzip archive.zip "docs/*.pdf"
# Extract, overwriting existing files without prompting
unzip -o archive.zip
# Extract, but NEVER overwrite existing files (skip conflicts silently)
unzip -n archive.zip
# Test archive integrity without extracting anything
unzip -t archive.zip
# Extract quietly (suppress the per-file listing normally printed)
unzip -q archive.zip
# View contents of a file INSIDE the archive without extracting it
unzip -p archive.zip path/inside/archive.txtzip supports compression levels 0 (no compression, fastest) through 9 (maximum compression, slowest):
zip -0 archive.zip file.txt # store only, no compression — fastest, largest
zip -1 archive.zip file.txt # fastest compression
zip -6 archive.zip file.txt # default balance (used if no level specified)
zip -9 archive.zip file.txt # maximum compression — slowest, smallest
# Practical trade-off check:
time zip -1 fast.zip bigfile.dat
time zip -9 small.zip bigfile.dat
ls -lh fast.zip small.zip-0 is particularly useful when you just want to bundle files without spending CPU on compression — for example, when the files are already compressed (video, images) and further compression would waste time for negligible size gain.
# Create a password-protected archive (prompts interactively — recommended)
zip -e archive.zip file.txt
# Non-interactive password (⚠️ visible in shell history and process list — avoid in scripts)
zip -P "mypassword" archive.zip file.txt
# Extract a password-protected archive
unzip archive.zip
# [archive.zip] file.txt password:
# Non-interactive extraction with password
unzip -P "mypassword" archive.zipImportant security caveat: the traditional ZIP encryption (ZipCrypto, used by the classic -e/-P flags in standard zip) is cryptographically weak and can be broken relatively easily with known-plaintext attacks — it should not be relied on to protect genuinely sensitive data. For real security, use AES-256 encryption, available via the -e flag combined with a build of zip that supports it, or more reliably via the 7z tool, or by encrypting the archive separately with GPG:
# Stronger alternative: encrypt with GPG after zipping
zip -r archive.zip myfolder/
gpg -c archive.zip # prompts for a passphrase, produces archive.zip.gpg
gpg -d archive.zip.gpg > archive.zip # decrypt later| Option | Description |
|---|---|
-r |
Recurse into directories |
-u |
Update — add new files, replace changed ones, skip unchanged |
-d |
Delete specified entries from an existing archive |
-m |
Move — delete original files after adding them to the archive |
-x PATTERN |
Exclude files matching a pattern |
-i PATTERN |
Include only files matching a pattern |
-e |
Encrypt entries, prompting for a password |
-P PASSWORD |
Encrypt with a password given on the command line (insecure for scripts) |
-0 to -9 |
Compression level (0=store, 9=max) |
-q |
Quiet — suppress normal output |
-v |
Verbose |
-j |
Junk paths — store just the file names, discarding directory structure |
-y |
Store symlinks as symlinks, not their targets |
-sf |
Show files that would be operated on, without doing anything |
--exclude=GLOB |
GNU-style exclude pattern (alternative syntax to -x) |
-T |
Test the archive's integrity after creating it |
-A |
Adjust self-extracting archive |
| Option | Description |
|---|---|
-l |
List contents without extracting |
-t |
Test archive integrity |
-d DIR |
Extract into a specific directory |
-o |
Overwrite existing files without prompting |
-n |
Never overwrite existing files |
-q |
Quiet mode |
-v |
Verbose listing (more detail than -l) |
-p |
Extract a file to stdout (pipe-friendly) |
-x PATTERN |
Exclude specific files from extraction |
-P PASSWORD |
Supply a password non-interactively |
-j |
Junk paths — extract files flat, ignoring stored directory structure |
-a |
Auto-convert text files' line endings for the current platform |
-C |
Case-insensitive filename matching |
-L |
Convert extracted filenames to lowercase |
-Z |
Invoke zipinfo-style listing (used internally by -l/-v) |
| Feature | .zip |
.tar.gz |
.7z |
|---|---|---|---|
| Universal OS support (no extra tools) | ✅ Windows/macOS/Linux all native | ❌ Needs 7-Zip installed everywhere | |
| Compression ratio | Moderate (per-file) | Better (whole-stream) | Best (LZMA2) |
| Random access to single files | ✅ Fast (Central Directory) | ❌ Must read sequentially from start | ✅ Fast |
| Preserves Unix permissions/symlinks reliably | ✅ Full fidelity | ✅ Full fidelity | |
| Strong built-in encryption | ❌ None (needs external gpg/openssl) | ✅ AES-256 by default | |
| Best use case | Cross-platform sharing, email attachments, general compatibility | Unix/Linux backups, preserving exact permissions/ownership | Maximum compression, archival storage |
# Same content, different formats — compare sizes:
zip -r archive.zip myfolder/
tar -czf archive.tar.gz myfolder/
7z a archive.7z myfolder/
ls -lh archive.zip archive.tar.gz archive.7zRule of thumb: choose .zip when the recipient's environment is unknown or non-technical; choose .tar.gz for Unix-to-Unix transfers where exact permissions/ownership matter; choose .7z when squeezing out the smallest possible size matters more than universal compatibility.
The original ZIP format has hard limits: 4 GB per file and 65,535 entries per archive, and a maximum archive size of 4 GB, due to 32-bit fields in the original 1989 specification. Zip64 is an extension that lifts these limits using 64-bit fields, and is used automatically by modern zip/unzip when needed:
# Archiving something over 4GB automatically triggers Zip64 extensions
zip -r bigarchive.zip huge_file_5gb.dat
# No special flag needed on modern Info-ZIP — it's automatic
# Force Zip64 format even for small archives (rarely necessary)
zip -fz archive.zip file.txtCompatibility caveat: very old ZIP tools (or minimal/embedded unzip implementations) may not understand Zip64 extensions and could fail on archives that exceed the original limits — worth checking when the recipient's tooling is unknown or aged.
| Command | Relation |
|---|---|
tar |
Alternative archiver, typically paired with gzip/xz; better for Unix permission fidelity |
gzip / gunzip |
Single-file compression, often used alongside tar instead of zip |
7z |
More modern format with stronger compression and encryption |
zipinfo |
Detailed listing of a ZIP archive's contents (what unzip -v calls internally) |
zipcloak |
Add/remove encryption on an existing (unencrypted) ZIP archive |
zipsplit |
Split a large ZIP archive into smaller volume pieces |
gpg |
Recommended for real encryption, layered on top of a zip archive |
file |
Identify whether a given file is actually a valid ZIP archive |
See also:
examples.md·edge-cases.md·interview-questions.md