Optimise PDF

Repair a damaged PDF

Recover readable content from a damaged PDF, or return a file that still opens.

Choose a file

Choose one unprotected PDF, up to 200 pages or 120 MB.

or drag and drop hereRuns on this deviceNo file upload

Repair a damaged PDF

Check a PDF with a damaged header or cross-reference structure, recover complete supported content when possible and leave the source file unchanged.

Return readable PDFs without rewriting them

The first pass uses the normal strict reader. If it opens the file and finds a complete usable page tree, the tool returns the original bytes instead of normalising or resaving the document.

Keeping the bytes unchanged preserves incremental history, metadata, signatures and specialised structures that a writer pass could disturb. It confirms that this reader can open the file, but it does not certify compatibility with every viewer.

Repair does not reduce file size, change the PDF version or reorder healthy internal structures. Use Compress PDF for size, Unlock PDF for authorised encryption and the editor for content changes.

Recover only complete document structures

Only a file rejected by the strict reader enters recovery. The recovery parser must find the complete page set, valid page boxes and a document structure that can be inspected before it writes a new copy.

The copy is built from reachable objects, which can remove broken unreachable fragments left by an interrupted download or faulty producer. A truncated page tree, ambiguous reference, missing page or impossible geometry ends with an error.

Recovery creates a separate candidate and leaves the source untouched. The candidate must pass the remaining structural and visual checks; opening successfully by itself is not enough for download.

Check document structure and page appearance

For a changed recovery, the source and final bytes are read independently. Checks cover page count, dimensions, rotation, forms, annotations, bookmarks, page labels, named destinations and embedded attachments.

Supported optional-content layers are compared by name, order, usage and display or print visibility. Text content and annotation behaviour are checked directly, and any detected difference stops the download.

Every source and result page is also rendered at the same geometry and compared visually. This separate check can find a shifted font, missing image or colour change that structural counts do not describe.

Leave signed and standards-bound files unchanged

A full rewrite changes byte offsets and cannot preserve a certificate-backed digital signature or document timestamp on a damaged source. Encryption, DocMDP permissions and XFA packages also carry behaviour that this recovery writer does not retain.

Tagged PDFs contain an accessibility tree with reading order and relationships, while PDF/A and PDF/X add conformance requirements. The tool therefore refuses to rebuild damaged files containing these features.

A signed or tagged file that opens normally follows the unchanged path and keeps its exact bytes. Damaged tagged, standards-bound or XFA files require a specialist application that can validate those structures.

Work within the repair limits

One repair job accepts one PDF up to 200 pages and 120 MB. Parsing, reachable-object rebuilding, structural inspection and page rendering run in this browser without sending the source, filename, text or attachments to a processing server.

Keep the damaged source after a successful repair. A recovered result is a separate normalised copy and cannot restore bytes that were absent from the selected file or replace signed evidence and archival provenance.

The output is unchanged when the strict reader accepts the source, rebuilt only after the supported structural and visual checks pass, and withheld when recovery cannot meet those conditions.

Questions
Which damaged PDFs can be repaired?

Recovery requires complete pages and supported structures. Damaged encrypted, signed, tagged, XFA, PDF/A and PDF/X files are not rebuilt.

What happens to digital signatures?

A readable signed PDF is returned byte-for-byte unchanged. A damaged signed PDF is not rebuilt because a rewrite would invalidate its signature.