How do I extract a ZIP file without uploading it?

Choose Open a ZIP, or drop the archive on the left pane. Its list comes from the table of contents at the end of the file, read in this browser tab, and the archive is never uploaded. Tick the files you want and choose Extract selected, or choose Extract all. Each file downloads on its own under its base name: the folders in its path are dropped, and a name that repeats is numbered, such as notes (2).txt. Every file is checked against the CRC-32 stored for it before it is saved.

How do I merge two ZIP files into one?

Choose Add a ZIP's contents for each archive, or open one on the left, choose Add all to new ZIP, then add the other. A name that is in both is listed as a name clash with two choices: Rename keeps both and numbers the later copy, such as notes (2).txt, and Replace keeps only the later copy. Names that differ only in upper and lower case count as a clash, because Windows and macOS treat them as one file by default. Make ZIP is refused until every clash is settled; when there are several, Rename all or Replace all settles them in one step. Folders travel in the paths of the files inside them, so an empty folder is not copied.

Can it open a password-protected ZIP?

No. An encrypted entry is listed with the method that locked it, ZipCrypto, WinZip AES or PKWARE strong encryption, and marked refused: this tool does not decrypt, so it never extracts that entry or copies it into a new ZIP. An archive whose table of contents is itself encrypted does not open at all, because not even its file names can be read.

Why is an entry marked rejected?

Its name could put the file outside the folder it is extracted to: it climbs out with ../, starts at the root of the disk with /, starts with a drive letter such as C:, holds a backslash, which Windows reads as a folder, or holds control characters, including the ones that reverse the text after them and can make a program look like a text file. Hidden characters are shown spelled out, such as . A rejected entry stays in the list, and it is never extracted or copied into a new ZIP. A name you type for a file in a new ZIP is checked the same way.

Is there a size limit?

Each job, one extraction or one new ZIP, expands at most 250 MB (262,144,000 bytes). The sizes declared for the files in the job are added up before anything is read, and a job over the limit is refused with the total it would reach. While a file unpacks, its bytes are counted as they arrive, and reading stops the moment they pass the size its header records, so an archive that understates its sizes is stopped as well. An archive that expands past 250 MB in total still opens, so you can extract a selection that fits. An archive that lists more than 50,000 entries, or whose table of contents is over 32 MB, is not opened.

Does it open RAR, 7z or TAR files?

No, ZIP only. A RAR, 7z, TAR, gzip, bzip2 or xz file is recognised from its first bytes and refused with a message that names its format. Files that are ZIP archives under another name, such as .docx, .xlsx, .epub and .jar, open and list their parts. Inside a ZIP, entries compressed with stored or deflate unpack; any other method, such as Deflate64, BZIP2, LZMA or Zstandard, is named and refused.

What do the compression levels do?

Level 0 stores each file as it is. Levels 1 to 9 deflate them, from fastest at 1 to smallest at 9, and 6 is the usual default. Files that are already compressed, such as JPEG photos, MP4 video or other ZIPs, shrink little at any level. The level is the only thing this page remembers in the browser.

ZIP Extractor and Creator

Open a ZIP to see every entry with its size, packed size, compression method and status, then extract the files you tick, or all of them. Make a new ZIP from files, a folder or other archives at a compression level from 0 to 9, and merge archives by adding the contents of each. Three checks come before anything is written: a name that appears twice in a new ZIP waits for you to pick Rename or Replace, a path that climbs out of the folder or starts at a root is listed and never extracted, and every job stops at 250 MB of expanded data, checked from the declared sizes before anything is unpacked. Archives and files are read in this tab and never uploaded.

Open a ZIP

no archive open

Drop a ZIP here

Or use Open a ZIP below. The list comes from the archive's own table of contents, read in this tab; nothing is uploaded.

Open a ZIP

new ZIP

empty

Make a new ZIP

Add files, a folder or a ZIP's contents below, or drop files here. Add two ZIPs' contents to merge them.

Add filesAdd a folderAdd a ZIP's contents

Entries: none open

Expanded: none

Rejected paths: none

Name clashes: none

ZIP name

Compression level

New ZIP: empty

Extracted files download one at a time under their own names: the folders in a path are dropped, and repeated names are numbered, such as notes (2).txt. Your browser may ask once before it saves several files.

Each job, an extraction or a new ZIP, expands at most 250 MB. Archives and files are read in this tab and never uploaded, and only the compression level is remembered in this browser.

Try an example

Archives written for this page and marked as examples. Each goes through the same reader as a ZIP of your own.

Two files named notes.txtmerging updates.zip into site-backup.zip
nothing is written until you pick rename or replace
Entries ../secret and /tmp/notes.txta climbing path and an absolute one
both rejected and listed; never extracted or copied into a new ZIP
An archive that expands past 250 MBsizes checked before reading
Extract all stops before anything is unpacked, with the size it would reach

Accuracy. It opens and makes ZIP files only: RAR, 7z, TAR and gzip files are named and refused. Every extracted file is checked against the CRC-32 stored for it, and an entry encrypted with ZipCrypto, WinZip AES or PKWARE strong encryption, or compressed with a method other than stored or deflate, stops with an error naming the method. Paths that climb out of the folder, start at a root or a drive letter, or hold backslashes or control characters are rejected, and each job expands at most 250 MB, checked from the declared sizes before reading and again as the bytes arrive.

Common questions

How do I extract a ZIP file without uploading it?: Choose Open a ZIP, or drop the archive on the left pane. Its list comes from the table of contents at the end of the file, read in this browser tab, and the archive is never uploaded. Tick the files you want and choose Extract selected, or choose Extract all. Each file downloads on its own under its base name: the folders in its path are dropped, and a name that repeats is numbered, such as notes (2).txt. Every file is checked against the CRC-32 stored for it before it is saved.
How do I merge two ZIP files into one?: Choose Add a ZIP's contents for each archive, or open one on the left, choose Add all to new ZIP, then add the other. A name that is in both is listed as a name clash with two choices: Rename keeps both and numbers the later copy, such as notes (2).txt, and Replace keeps only the later copy. Names that differ only in upper and lower case count as a clash, because Windows and macOS treat them as one file by default. Make ZIP is refused until every clash is settled; when there are several, Rename all or Replace all settles them in one step. Folders travel in the paths of the files inside them, so an empty folder is not copied.
Can it open a password-protected ZIP?: No. An encrypted entry is listed with the method that locked it, ZipCrypto, WinZip AES or PKWARE strong encryption, and marked refused: this tool does not decrypt, so it never extracts that entry or copies it into a new ZIP. An archive whose table of contents is itself encrypted does not open at all, because not even its file names can be read.
Why is an entry marked rejected?: Its name could put the file outside the folder it is extracted to: it climbs out with ../, starts at the root of the disk with /, starts with a drive letter such as C:, holds a backslash, which Windows reads as a folder, or holds control characters, including the ones that reverse the text after them and can make a program look like a text file. Hidden characters are shown spelled out, such as <U+202E>. A rejected entry stays in the list, and it is never extracted or copied into a new ZIP. A name you type for a file in a new ZIP is checked the same way.
Is there a size limit?: Each job, one extraction or one new ZIP, expands at most 250 MB (262,144,000 bytes). The sizes declared for the files in the job are added up before anything is read, and a job over the limit is refused with the total it would reach. While a file unpacks, its bytes are counted as they arrive, and reading stops the moment they pass the size its header records, so an archive that understates its sizes is stopped as well. An archive that expands past 250 MB in total still opens, so you can extract a selection that fits. An archive that lists more than 50,000 entries, or whose table of contents is over 32 MB, is not opened.
Does it open RAR, 7z or TAR files?: No, ZIP only. A RAR, 7z, TAR, gzip, bzip2 or xz file is recognised from its first bytes and refused with a message that names its format. Files that are ZIP archives under another name, such as .docx, .xlsx, .epub and .jar, open and list their parts. Inside a ZIP, entries compressed with stored or deflate unpack; any other method, such as Deflate64, BZIP2, LZMA or Zstandard, is named and refused.
What do the compression levels do?: Level 0 stores each file as it is. Levels 1 to 9 deflate them, from fastest at 1 to smallest at 9, and 6 is the usual default. Files that are already compressed, such as JPEG photos, MP4 video or other ZIPs, shrink little at any level. The level is the only thing this page remembers in the browser.

It opens and makes ZIP files only: RAR, 7z, TAR and gzip files are named and refused. Every extracted file is checked against the CRC-32 stored for it, and an entry encrypted with ZipCrypto, WinZip AES or PKWARE strong encryption, or compressed with a method other than stored or deflate, stops with an error naming the method. Paths that climb out of the folder, start at a root or a drive letter, or hold backslashes or control characters are rejected, and each job expands at most 250 MB, checked from the declared sizes before reading and again as the bytes arrive.