Guides

What is PDF/A?

PDF/A is a version of PDF made for keeping documents for a long time. A PDF/A file should open and look exactly the same in ten, twenty or fifty years, in software that does not exist yet, without depending on anything outside the file itself.

If a court, a government portal, a university or an archive has asked you for a PDF/A file, this guide explains what that means, which version to choose, and how to convert and check a file with A2ZDocs.

The standard: ISO 19005

PDF/A is an international standard, ISO 19005. It takes ordinary PDF and adds rules: some features become compulsory, and others are forbidden because they could make a file look different later or depend on something that might disappear. It has been published in parts, each based on a newer version of PDF:

  • PDF/A-1 (ISO 19005-1, 2005), based on PDF 1.4. The strictest part: no transparency, no layers and no attached files.
  • PDF/A-2 (ISO 19005-2, 2011), based on PDF 1.7. It allows transparency, layers, JPEG 2000 images and attachments, as long as the attachments are PDF/A files themselves.
  • PDF/A-3 (ISO 19005-3, 2012). The same as PDF/A-2, except that files of any type can be attached, such as the spreadsheet or data a document was made from.

A newer part, PDF/A-4 (ISO 19005-4, 2020), is based on PDF 2.0. A2ZDocs works with parts 1, 2 and 3.

Conformance levels: b, u and a

Each part comes in levels, shown by the letter after the number, as in PDF/A-2b:

  • b (basic): the pages will look the same in future. This is the minimum level.
  • u (Unicode): level b, plus every character of text is mapped to Unicode, so the text can be reliably searched and copied. It exists in PDF/A-2 and PDF/A-3.
  • a (accessible): adds a tagged structure that records headings, paragraphs, lists, reading order and descriptions of pictures, so the document works with screen readers. In PDF/A-2 and PDF/A-3 it includes everything in level u as well.

A stricter level is not better for every purpose. If you are not told which one to use, PDF/A-2b is a sensible choice, and it is what A2ZDocs selects unless you pick another.

Why PDF/A matters

Ordinary PDFs can depend on things outside the file: fonts installed on a computer, links to other files, scripts that run when the file opens. Any of these can change or vanish. PDF/A removes those dependencies, which is why it is asked for wherever documents must be kept and trusted for years:

  • National, state and university archives, and libraries that keep digital collections.
  • Courts and electronic filing systems that ask for submissions in PDF/A.
  • Government portals and tenders, and universities collecting theses and dissertations.
  • Organisations storing records they are required to keep for many years.

Always follow the exact instructions of whoever asked. If they name a part and level, use that one.

What PDF/A requires and forbids

The full rules run to many pages, but the main ones are:

  • Embedded fonts. Every font used to show text must be inside the file, so the text looks the same on any computer.
  • Defined colour. The file carries a colour profile, called an output intent (for example sRGB), so colours can be reproduced on any screen or printer.
  • No encryption. A PDF/A file cannot have a password or permission restrictions, because a lost password would lock the archive.
  • No external content. Everything needed to show the file must be inside it, with no references to outside files.
  • No JavaScript, and no actions that launch programs or run code. Audio and video are not allowed either.
  • Standard metadata. The file must contain XMP metadata that declares which part and level it meets.
  • Fixed appearances. Comments and form fields must have a fixed appearance and be set to print.
  • In PDF/A-1 only: no transparency, no layers, no JPEG 2000 images and no attachments.

Converting a PDF to PDF/A with A2ZDocs

  1. Open PDF to PDF/A and choose one or more PDFs. They are processed in your browser and not uploaded. If a file has a password, remove it first with Unlock, because PDF/A files cannot be encrypted.
  2. Choose the PDF/A level you were asked for. If no one has said, choose PDF/A-2b.
  3. Leave "Flatten form fields" ticked, so filled-in answers become a fixed part of the page.
  4. For PDF/A-1 only: if the document uses transparency or JPEG 2000 images, tick "Flatten transparent pages to images". Without it, the conversion stops and tells you which pages are affected. Flattened pages become images, so their text can no longer be selected.
  5. Select Convert to PDF/A. Your new file ends in _pdfa.pdf. Read the notes that come with it.

The converter adds what PDF/A requires: XMP metadata that matches the document's properties, an sRGB colour profile, a file identifier and the right file header. It removes what PDF/A forbids, such as JavaScript, actions that run automatically, references to outside files and attachments the chosen part does not allow. It also recompresses data stored in the old LZW format, which PDF/A does not allow, and gives comments and form fields a fixed appearance. Finally, it checks the result and tells you whether any problems remain.

What the converter cannot fix

  • Missing fonts. If the PDF uses a font it does not contain, the website cannot add it, so the result is not fully PDF/A compliant and the notes say so. The best fix is to save as PDF/A directly from the program that made the document; many word processors and design programs offer a PDF/A option when saving or exporting a PDF.
  • Untagged files at level a. The a levels need a tagged PDF. If your file has no tags, A2ZDocs tells you and suggests level b or u instead. Tags have to come from the program that created the document.
  • Text without Unicode. If a font does not record which characters its shapes stand for, level u is not possible. The file is then saved as level b instead, with a note.
  • CMYK colour. Print-ready files that use CMYK colour may not be fully covered by the sRGB profile A2ZDocs adds; the notes will tell you if so.

If your document is a scan and the portal also wants searchable text, run OCR before converting.

Checking a file with Check PDF/A

Check PDF/A tests files against the PDF/A rules on your device and writes a report for each one. You can check against the level a file says it meets (PDF/A-2b if it says nothing) or choose a level yourself. The report lists problems such as fonts that are not embedded, missing or mismatched metadata, a missing colour profile, encryption, transparency in PDF/A-1, forbidden actions or JavaScript, unsuitable attachments and comments without a fixed appearance, plus warnings worth a look.

It is useful for finding out why a portal rejected your file, and for checking a converted file before you submit it.

An honest note about validation

The checks in A2ZDocs cover the most common reasons files fail PDF/A, but they are a practical preflight, not a certified validator, and passing them is not an official certification. For level a, for example, they check that the file is tagged, not that the tags are correct. The widely used open-source validator for PDF/A is veraPDF. If an organisation needs formal proof that a file conforms, validate it with veraPDF or with whatever tool they specify.

Tools in this guide

All guides