The original ZIP format stores sizes and offsets as 32-bit numbers, which was plenty in 1989 and is not today. ZIP64 is the extension that lets a ZIP hold files over 4 GB, grow beyond 4 GB in total and contain more than 65,535 entries. Most of the time it works without anyone noticing, but it explains some puzzling errors with large archives.
The classic ZIP limits
| Field | Width | Limit |
|---|---|---|
| Size of one file (compressed or uncompressed) | 32-bit | 4,294,967,295 bytes (4 GiB minus 1 byte) |
| Offset of a file within the archive | 32-bit | Each file must start within the first 4 GiB |
| Size and offset of the central directory | 32-bit | 4 GiB |
| Number of entries | 16-bit | 65,535 |
In practice that means no single file over 4 GB, no archive much over 4 GB, and no more than 65,535 files and folders in one classic ZIP.
How ZIP64 works
ZIP64 was added to the ZIP specification in 2001. It keeps the familiar structure and adds 64-bit versions of the fields that can overflow:
- When a value does not fit, the classic field is set to a marker (
0xFFFFFFFF, or0xFFFFfor counts) and the real value goes into a ZIP64 extra field (header ID0x0001) attached to the entry. - A ZIP64 end of central directory record holds 64-bit entry counts and the size and position of the central directory.
- A ZIP64 end of central directory locator, placed just before the classic end record, tells readers where to find it.
64-bit fields allow sizes up to 16 exabytes, so with ZIP64 the limit becomes your storage, not the format.
When ZIP64 is used
A well-behaved tool switches to ZIP64 only when needed: when a file or the archive passes 4 GB, or the number of entries passes 65,535. Some streaming tools add ZIP64 fields to every file because they cannot know sizes in advance. That is valid, but it can confuse older software.
Create ZIP and the other Aback ZIP tools add ZIP64 records only when a value actually overflows, so ordinary archives stay as compatible as possible.
Which tools support ZIP64
| Tool | ZIP64 support |
|---|---|
| 7-Zip, WinZip, WinRAR, PeaZip, Bandizip, Keka | Yes |
| Windows File Explorer (Windows 10 and 11) | Reads ZIP64 archives; if a very large archive fails there, 7-Zip is the usual fallback |
| macOS Archive Utility | Yes |
Info-ZIP zip 3.0 and unzip 6.0 | Yes; older versions (zip 2.x, unzip 5.x) do not |
Java 7+, Python zipfile, .NET | Yes |
| Older embedded devices, firmware updaters, legacy libraries | Often not; keep archives under 4 GB and 65,535 entries for these |
ZIP64 and FAT32 drives
FAT32, still common on USB sticks and SD cards, cannot store any single file of 4 GB or more, whatever format it is in. If a large archive will not copy to a drive, the drive's file system is the problem, not ZIP64.
- Reformat the drive as exFAT or NTFS. This erases the drive, so copy anything you need off it first.
- Or split the archive into parts under 4 GB that each open on their own.
ZIP64 in the browser
JavaScript numbers are exact up to 9,007,199,254,740,991, about 9 petabytes, so 64-bit ZIP fields can be handled safely as long as the code checks that bound. Aback ZIP does, and reports an error rather than misreading anything larger.
Files are read in slices rather than loaded whole, and ZIP Viewer reads only the central directory, so it can show whether an archive uses ZIP64 without loading all of it. The practical ceiling for creating or extracting very large archives is your device's memory and storage, and phone browsers have much less room than desktop browsers.