PDF/A is an ordinary PDF with a set of promises attached. Every font is inside the file. Colours are defined so they do not depend on one printer or screen. Nothing is encrypted, nothing runs a script, nothing points to a file somewhere else. The aim is simple: a document that opens and looks the same in thirty years as it does today. This page converts a PDF to PDF/A-2b on our servers, makes scanned pages searchable on the way and checks every result against the standard.
Who asks for PDF/A, and why
- Archives and records management. Records kept for 10 years or forever should not depend on a font that happens to be installed today. Records policies often name PDF/A as the format for final versions.
- Public sector. Many administrations store final documents as PDF/A, and some require a text layer that can be searched, which rules out a plain scan. If an authority asks for PDF/A, its guidelines usually name the part (PDF/A-1, -2 or -3).
- Courts. Rules differ by country and court. E-filing portals often accept PDF/A alongside ordinary PDF without requiring it. Check the rules of the court you file with before you convert.
- Your own archive. Contracts, certificates, scanned letters: if you want to find them by searching their text years from now, a searchable PDF/A is a sensible container.
What our converter does with your PDF
- Pages that already contain text keep their content. The file is rewritten to meet the PDF/A-2b rules, but text, images and layout stay as they were.
- Scanned pages (a photo of a page with no text in it) go through text recognition (OCR). The recognised text is placed as an invisible layer under the image, so the page looks exactly like the scan, yet you can search it and copy text from it. Recognition covers documents up to 30 pages; a longer scan is converted to PDF/A without it. We read English, German, Czech and Slovak; Cyrillic, Greek, Arabic, Hebrew, Japanese, Chinese and Korean are detected from the script of the first page.
- Elements PDF/A forbids, such as JavaScript, embedded audio and video, or encryption, cannot be part of the result.
- Every result is checked by an independent validator against the rules of ISO 19005-2, level B. The download step shows either “Verified: meets PDF/A-2b” or the number of rules the file does not meet. You know before you submit it, not when an archive rejects it.
The result is a normal file with the .pdf extension. Any PDF reader opens it; some readers show a bar saying the document is in PDF/A mode and opened read-only. That is expected.
PDF/A-1, 2, 3, a, b, u: what the letters mean
The number is the part of ISO 19005, the letter is the conformance level. PDF/A-2 (ISO 19005-2, published 2011) builds on PDF 1.7 and, unlike PDF/A-1, allows transparency, JPEG 2000 compression and smaller files through compressed object streams.
| Variant | Based on | What it adds |
|---|---|---|
| PDF/A-1 | PDF 1.4 (ISO 19005-1:2005) | The original. No transparency, no attachments. |
| PDF/A-2 | PDF 1.7 (ISO 32000-1) | Transparency, JPEG 2000, attachments only if they are PDF/A themselves. |
| PDF/A-3 | PDF 1.7 | Attachments of any format, used for e-invoices with embedded XML. |
| Level b | The page looks right. This is what we produce. | |
| Level u | Level b plus Unicode for all text. | |
| Level a | Full logical structure (tags), for accessibility. |
Check which part a portal or court names: some accept several parts, others ask for one in particular. We do not produce PDF/A-1, PDF/A-3 or level a at the moment. The Library of Congress description of the PDF/A family lists all the restrictions in detail.
Common problems when converting PDF to PDF/A
A signed PDF is rejected
Any change to a signed PDF breaks the signature, and conversion to PDF/A is a change. We therefore do not convert digitally signed PDFs and tell you why next to the file. Convert the unsigned document to PDF/A first, then sign the PDF/A.
PDF with a password
Encrypted PDFs cannot be converted, and PDF/A does not allow encryption anyway. If you are entitled to, open the file, enter the password and save a copy without protection.
The scan is still not searchable
Check the page count: scans over 30 pages are converted without recognition. Split the file and convert the parts. Recognition also suffers from low-resolution scans, faxes, skewed pages and handwriting, so search for a word you can see before relying on it.
The check reports failed rules
Some source files are damaged or built in unusual ways. If the result does not conform, try printing the original to a new PDF in your viewer and convert that copy, or export again from the program that created it.
Other ways to get a PDF/A
If the original is a Word document, Word on Windows can save PDF/A directly: Save As, choose PDF, then Options and tick “ISO 19005-1 compliant (PDF/A)”, as Microsoft describes. Word makes PDF/A-1 and does not tell you whether the file passed a check. For Word files you can also use DOCX to PDF/A here.
Our converter helps when all you have is a PDF or a scan, when you need the text layer, or when you have a batch: up to 10 files at once, each up to 100 MB, without signing up. The uploaded PDF is deleted right after conversion, the result after 14 days. Need an editable document instead? Use PDF to DOCX. Format background: PDF.