ExcelAnalyzerClient-side file inspector

Excel corrupted file checker

Verify that no file leaves your browser

Local check

What Excel means by unreadable content

The message “Excel found unreadable content” is broad. Sometimes the ZIP package is truncated because a download stopped. Sometimes workbook.xml lists a sheet whose file is missing. Sometimes one part is not well-formed XML: a tag was cut off, or two parts were merged by a bad sync. Excel’s repair then removes whatever it cannot understand, which can be a chart, a pivot, or a whole sheet. This checker names the part and the kind of failure before you let Excel discard it.

What is tested

The file must start with a ZIP signature and must have a readable central directory. [Content_Types].xml, the root relationships, and the workbook part are required. Each sheet listed in workbook.xml must have its worksheet part. XML parts under the package, up to 15 MB uncompressed each, are parsed for well-formedness. That means tags match and the file is not cut off mid-tag. It is not a full schema validation, so a part can be well formed and still contain a value Excel dislikes. Encrypted packages are reported directly: the encrypted blob cannot be inspected, and the password is never requested.

The repair steps live in the guide on what “unreadable content” usually means.

What to do with a named part

If the package is not a ZIP, get a new copy from the sender. Repairing a half-downloaded file is not realistic. If a sheet part is missing, Excel’s repair will drop that sheet. If you have an older copy, compare sheet names before you accept the repaired file. If a single XML part is not well formed, Excel sometimes recovers after you let it repair, and sometimes it removes more than the broken part. Save the repaired workbook under a new name. Do not overwrite the only copy. Then run this check on the repaired file. A clean result means the package structure is whole again. It does not mean every number is the number you had last Tuesday.

What this is not

This is not a forensic recovery lab. It will not reconstruct a sheet from a truncated ZIP. It will not bypass a password. It will not turn an .xls file into an .xlsx file. Legacy .xls is refused up front because it is a different format, not because it is corrupt. A modern file that is actually an old .xls with a renamed extension is refused with the same guidance.

After the package is in order

Corruption checks do not replace the other checks. A file can be well formed and still full of #REF!, external links, and a phantom used range. Those are consistent files that are still wrong. Once this page is clean, use the error finder and the size checker on the repaired copy.

Also check

Questions people ask before trusting the result

Should I click Yes when Excel offers to repair?

Only on a copy. Let Excel write the repaired workbook to a new file name, then compare it with what this page reported.

Why was a large pivot cache not fully scanned?

Parts over 15 MB uncompressed are not loaded into an XML parse. The size is still measured. A skip is a warning about coverage, not a claim that the part is corrupt.

The file opens in Excel. Why is it flagged?

Excel tolerates some defects and repairs them quietly on save. The flag names the defect so you can see it before a later save changes the file.

Can I fix the XML myself?

Only if you already know the part and you keep a backup. Editing workbook XML by hand is easy to make worse. Excel’s repair is safer for most people.

Do you keep the broken file to debug it?

No. The file is read in your browser and dropped when you reset the page. There is no server copy.

Last updated 5 November 2026. Initial release of this audit.