Excel found unreadable content: Repair a Corrupt Workbook and Recover Data

Fix it now

Excel opened the container but could not load every part inside it, and is offering to salvage what it can. Sometimes the workbook really is damaged; sometimes it is intact and simply is not the format its extension claims. Work on a copy either way.

1. Copy the file to a new folder and leave the original untouched. Everything below happens to the copy.
2. Run Excel’s own repair: on the File tab choose Open, select the file, click the arrow next to the Open button, and click Open and Repair. Choose Repair to recover as much data as possible.
3. If Repair cannot do it, run the same route again and choose Extract Data.
4. If Excel says the format or extension is not valid, check what the file really is by opening it in a text editor. A file that reads as HTML or as comma-separated text is not a workbook and no repair tool will help.
5. If the file came from OneDrive or SharePoint, try version history first: open the file location in a browser and restore or download an earlier version.
6. If the repair produced a log listing removed records, note which sheets and parts it names. That is what you need to rebuild.

Repair rewrites the file. Never run it against your only copy, and keep the copy you started with until you are satisfied with the result.

If the repaired workbook opens with no prompt and the numbers check out, you are done. The next section covers why a workbook stops being readable and what else Microsoft documents for recovering one.

Why it happens

A modern workbook is a container holding separate parts: one for each worksheet, plus parts for shared strings, styles, defined names, the calculation chain and anything else the file uses. Excel validates each part as it loads. If a part is truncated, malformed, or refers to something that is not there, Excel cannot load it as written and offers to repair rather than failing outright.

Repair works by dropping what it cannot parse and keeping the rest, which is why the repair log is far more interesting than the message itself. The log names the parts and record types it removed, so you know whether you lost one formula on one sheet or the whole style table. A workbook that repairs with a short log is usually fine to carry on with. One with a long log should have its data extracted into a fresh file rather than being trusted.

The second family of causes has nothing to do with damage. Applications that export what they call Excel files very often write an HTML table or delimited text and give it a spreadsheet extension. Excel notices the mismatch and complains about it, and no amount of repairing will change a file that was never a workbook. Files truncated by an interrupted download, a full disk, or a security product that quarantined part of the write produce a similar complaint from a different direction: the container is short, so the parts inside it do not add up.

Microsoft documents more recovery routes than most people know about, and they are worth trying in order before reaching for anything you have to buy. Reverting to the last saved version, saving in SYLK format, setting calculation to manual before opening, and pulling the data out with external references are all published methods, and between them they cover cases where Open and Repair alone gives up.

Full reference

Which problem you have

Microsoft’s documented recovery routes

Reading the repair log

The log Excel writes after a repair names each part it dropped and what kind of record it was. A line naming a formula from one worksheet part tells you one sheet lost calculation and everything else survived. A line naming shared strings or styles is more serious, because those are shared across the whole workbook. Check the named sheets against a known-good source before you rely on the file for anything, and if the log runs to more than a handful of lines, extract the data into a fresh workbook rather than continuing with the repaired one.

The messages themselves

None of the message strings in this article has a documented meaning published by Microsoft. That does not make them meaningless – they are Excel telling you what it did – but it does mean there is no vendor page to check a particular wording against, so read them as descriptions of the action offered rather than as codes to look up.

Preventing the next one

• Keep working files in OneDrive or SharePoint so version history exists. It is the single fastest recovery route and it costs nothing extra on a licence you already have.
• Avoid editing workbooks directly over an unreliable network link; save locally and copy up.
• Close Excel properly rather than powering off, which is where most truncated saves come from.
• Treat exports from other systems as imports to be checked rather than as workbooks to be trusted.
• Where a workbook is business-critical and large, keep a periodic copy outside the live location, because version history helps only if the file is where version history is kept.

Every code this article covers

Confirm the fix worked

1. The repaired workbook opens with no prompt and no repair offer.
2. A full recalculation produces no unexpected errors in cells that previously held values.
3. Every sheet named in the repair log has been checked against a known-good source or an earlier version.
4. Saved as a new file, closed and reopened, it is stable rather than repairing itself each time.
5. The original copy you set aside can be discarded, which is the point at which the job is finished.

Questions people ask about this