PDF Repair Tool — Fix a Corrupted PDF
Diagnose and repair a damaged PDF in your browser. Tells you exactly what is wrong before it tries anything.
Reading the file…
How to use this tool
- Select the damaged PDF, or drag it onto the box above.
- It is checked first — header, end-of-file marker, cross-reference pointer, encryption and object count — and the findings are shown before anything is altered.
- Repair runs automatically: a lossless structural rebuild first, page-by-page recovery only if that fails.
- Download the repaired file. The panel tells you which method was used.
Most PDF corruption is structural rather than missing data. A PDF ends with a cross-reference table listing where every object sits; when that breaks, readers refuse a file whose contents are almost entirely intact. Rebuilding it is genuinely possible, and it is lossless.
PDF Repair Tool — Fix a Corrupted PDF — the specification
| Diagnoses | Header, end-of-file marker, cross-reference pointer, encryption, object count |
|---|---|
| Repair 1 | Structural rebuild — lossless, text stays selectable |
| Repair 2 | Page-by-page recovery if the structure is beyond rebuilding |
| Cannot fix | Genuinely missing bytes — a truncated download |
| Limit | 100 MB |
PDF Repair Tool — Fix a Corrupted PDF: what can go wrong
Most failures here are not about the document itself but about how it was produced. The ones that actually come up:
- The download was interrupted, so bytes are actually missing. Nothing can invent them — download the file again from the source, which does fix this.
- The file is password-protected. Remove the password first; we do not bypass encryption.
- The file is not a PDF at all. A renamed Word document or image shows up as "no PDF header".
- The file is 0 bytes. There is nothing inside it.
How it works
The file is diagnosed before anything is changed. The raw bytes are checked
for the %PDF- header at the start, the %%EOF marker at
the end, a startxref pointer, an /Encrypt dictionary,
and how many PDF objects are actually present. Those five checks distinguish the
common failures from each other, and you see the findings before repair runs.
Repair then makes two attempts, in order. The first is a structural rebuild: the document is parsed with lenient settings, its objects are copied into a fresh document, and a clean cross-reference table is written. This is lossless — text stays selectable, images keep their quality — and it fixes the most common cause of a PDF that will not open, which is a broken index rather than missing content.
If the structure is too damaged for that, the fallback renders each page it can reach and assembles those images into a new PDF. This recovers pages that would otherwise be lost, but it flattens them: text becomes part of an image. The result panel tells you plainly which method was used, because the difference matters to what you can do with the file afterwards.
One thing repair cannot do is invent data. If a download was interrupted, the missing bytes are not anywhere in the file and no tool can reconstruct them.
A measured run of this tool
Not an estimate. This page’s own code, run against a documented reference file.
Input: 2.92 MB. Output: 2.92 MB.
The file size barely moved — up 0.0%. That is the point: this operation edits the PDF’s structure and copies the existing page objects rather than re-rendering them, so text stays selectable and images keep their original quality. A tool that re-encodes would show a much larger change here, and you would be paying for it in quality.
Who needs PDF Repair Tool — Fix a Corrupted PDF
A PDF that will not open, opens blank, shows "damaged file" in Acrobat, or was recovered from a failing drive or an interrupted transfer.
PDF Repair Tool — Fix a Corrupted PDF — frequently asked questions
Can a truncated PDF be recovered?
Only partly, and only the part that survived. If the download stopped halfway, the rest of the data does not exist anywhere in the file — no tool can reconstruct it. What repair can do is salvage the pages that did arrive. If you still have access to the source, downloading again is a better fix than any repair tool.
Will repair lose my text?
Not on the first pass. A structural rebuild copies the existing objects, so text stays selectable and images keep their quality. Only if the structure is unrecoverable does it fall back to rebuilding pages as images, and it tells you when that happens.
Why does my PDF open in one program but not another?
Readers differ in tolerance. Chrome and pdf.js rebuild a broken cross-reference table on the fly; Acrobat and many print systems refuse. Repairing writes a clean table so every reader accepts it.
Is my file uploaded?
No. Everything runs in your browser — which matters here, because damaged files are often the ones you most want back and least want to send to a stranger.
Is this tool really free?
Yes. No account, no trial period, no watermark on the output and no cap on how many times you use it. The site is funded by advertising.
Do I need to install anything?
No. There is nothing to download and no extension to add — it runs in the browser you are reading this in.
How a PDF breaks, and what a rebuild actually does
A PDF is a set of numbered objects plus a cross-reference table saying which byte each object starts at. Nearly all corruption is damage to that table rather than to the objects: the content is still present, but the index pointing at it is wrong, so the reader gives up.
That is why repair works as often as it does. A rebuild ignores the broken table, scans the whole file for object headers, works out where each one really begins, and writes a fresh table. Nothing in the page content is altered — it is re-indexed, not re-created.
When that fails, the fallback is page-by-page recovery: read whatever pages parse and write them to a new document, accepting the loss of any that do not. You get a smaller, working file instead of a complete, unopenable one.
What cannot be recovered is content that was never written. A download that stopped at 60% is missing 40% of its objects, and no amount of re-indexing invents them.
Related tools
From other categories
Background reading: How to Compress a PDF Without Ruining It
Your PDF never leaves this device. Contracts, statements and applications are processed entirely in the browser — open the Network tab while you use the page and you will see no upload (how that works).
Disclaimer. Upload limits differ between portals and are often lower than the published figure once the whole application is counted. Check the limit shown on the upload screen itself. Full disclaimer.