How do I recover files from a damaged ZIP archive?

First let a tolerant tool like 7-Zip extract every readable member. Then rebuild the archive with the repair ZIP tool or zip -FF, which scans the local headers directly and recovers members that a broken index was hiding. Re-source only the files that were truncated or failed their checksum.

Can I recover just some files from a broken ZIP?

Yes. A ZIP stores each file as an independent member, so a single damaged entry does not block the others. Recovery tools pull out the healthy members even when part of the archive is unreadable, which is why partial recovery of most files is a common and successful outcome.

What can I do with a CRC-failed entry?

A CRC error means that file's data was physically altered, so a rebuild cannot restore it. Get a clean copy of that one file from the source, or force-extract the partial data if a slightly damaged version is still usable. Meanwhile, recover every clean member normally.

Why can I only recover part of my ZIP archive?

The archive was almost certainly truncated during download or transfer, so it is physically shorter than it should be. Members before the cut-off point recover fully; members after it were never written to disk. Recovering the surviving files is the best possible result, and a fresh download fills the gap.

Does rebuilding a ZIP change my files?

No. A rebuild only reconstructs the archive's index around the existing members; it does not alter the recovered files themselves. Working on a copy means the original stays intact, so a rebuild can only add to what you recover, never take anything away.

Recovering Files from a Damaged ZIP Archive

Recovering files from a damaged ZIP archive is usually more successful than the scary error message suggests, because a ZIP is not one indivisible object. It is a collection of separately stored files, and damage to one part rarely destroys the rest. When an archive reports a broken central directory or a bad checksum, the practical question is not "is the file dead?" but "which entries can I still pull out?" This guide explains how member recovery works, how to extract everything readable, and what to do about the entries that will not come back cleanly.

To recover intelligently, you need a mental model of the format. Each file inside a .zip is stored as an independent member: a local header that names it and records its compression method and a CRC checksum, followed by the compressed data. Only after all the members does the archive place its central directory and end-of-central-directory (EOCD) record, the index that ties everything together. Because members stand on their own, a tool can often recover them one at a time even when the shared index is gone.

How Member Recovery Works

Recovering files from a damaged ZIP archive relies on that independence. Instead of trusting the broken index at the end, a recovery pass scans the archive from the beginning, looking for each local header signature in turn. Every time it finds one, it reads the entry's name and compression details, then decompresses the data that follows. Entry by entry, it reconstructs the file list directly from the members themselves, ignoring the damaged directory entirely.

This is exactly what Info-ZIP's zip -FF does on the command line and what the repair ZIP tool does in your browser. The rebuilt archive gets a fresh, valid central directory wrapped around whichever members survived, so it opens normally afterward. The insight worth remembering is that a missing index is a navigation problem, not a data-loss problem, and navigation can be reconstructed.

Step One: Try to Extract What Is Readable

Before any formal repair, see how much your existing tools can already pull out. Often more of the archive is reachable than the first error implied.

Open with 7-Zip. 7-Zip is unusually tolerant of damage. Right-click the archive, choose Open archive, and it frequently lists the members even when other tools refuse. Select what you can see and extract it.
Extract entry by entry. If extracting everything at once fails on one bad file, pull out the healthy members individually so a single damaged entry does not block the rest.
Note which files fail. Every tool that flags a failure is giving you a map of exactly which members are damaged and which are clean.

If this rescues everything you need, you are done, no rebuild required. If some members are unreachable because the index is broken, move on to a rebuild that scans the local headers directly.

Step Two: Rebuild to Recover Unreachable Members

When the damage is a broken central directory, extraction tools may only show part of the archive, because they still depend on that index to find entries. A rebuild fixes this by scanning past the damage.

Copy the damaged .zip so the original is safe.
Upload it to the repair ZIP tool, or run zip -FF broken.zip --out fixed.zip from a terminal.
The tool walks the local headers from the front, recovers each readable member, and writes a fresh index around them.
Open the rebuilt archive and extract the members that were previously unreachable.

After a rebuild, files that the broken index had hidden usually reappear, because the new index points straight at the members that were there all along.

Understanding Partial Recovery

Partial recovery is the realistic outcome when an archive was truncated, meaning it is physically shorter than it should be. This happens when a download or transfer stops early. The members near the front of the file arrived complete and recover perfectly; the members near the end are cut off or entirely absent.

In this situation a recovery pass gives you back the complete members and stops where the data runs out. Recovering nine of ten files from an archive that would not open at all is a real success, not a failure. What matters is understanding the boundary: everything before the truncation point is recoverable, and everything after it was never written to disk. No tool can recover data that never arrived, so for the missing tail your only route is a clean copy from the source.

Handling CRC-Failed Entries

Each member carries a CRC checksum, a fingerprint of its original uncompressed contents. During extraction, the unzip program decompresses the data and compares it against that fingerprint. A CRC error means the two do not match, so the member's data was physically altered after the archive was built, usually by disk decay, a bad transfer, or tampering.

A CRC-failed entry is a different problem from a missing index. The member is present and its structure is fine, but its bytes are wrong. A rebuild cannot correct this, because it can only reorganise data, not regenerate the exact original contents. Your options for that specific file are:

Get a fresh copy. If the archive came from a download, cloud drive, or colleague, a clean copy of that one file is the complete fix.
Extract the data anyway. Some tools let you force extraction past a CRC error, giving you a partial or slightly damaged file. For a large media file or document, imperfect data can still be usable; for anything that must be exact, it usually is not.
Recover the rest. A CRC failure on one member does not condemn the others. Recover every clean file first, then deal with the failed one separately.

A Sensible Recovery Order

Putting it together, work from least effort to most:

Extract what opens. Let 7-Zip or your usual tool pull out every readable member first.
Rebuild the archive. Use the repair ZIP tool or zip -FF to recover members that a broken index was hiding.
Re-source the gaps. For truncated tails and CRC-failed entries, fetch a clean copy of just those files from the original source.

This order recovers the maximum amount of data with the least wasted effort, and it always leaves your original untouched. To understand why the damage happened in the first place, and how to avoid a repeat, see why ZIP files get corrupted and how to prevent ZIP corruption. For the full repair walkthrough, our guide on how to repair a corrupted ZIP file covers every option.

Conclusion

Recovering files from a damaged ZIP archive works because the format stores each file as an independent member with its own header and checksum. Extract everything readable first, rebuild the archive to recover members a broken index was hiding, and re-source only the truncated or CRC-failed files that cannot be reconstructed. A partial recovery that returns most of your files is a genuine win. Copy your archive, run it through the repair ZIP tool, and rescue everything that survived.