PDF/A Explained: Levels, Rules and When You Need It

Open a PDF made today in twenty or thirty years and there is a fair chance it will still look right. A fair chance is not good enough for land records, court files, medical histories or company archives. PDF/A exists to turn that chance into something much closer to a promise, by stripping a PDF down to the parts that can be reproduced reliably without depending on anything outside the file.

What PDF/A is

PDF/A is an ISO standard (ISO 19005) that defines a restricted subset of PDF for long-term archiving. It is not a different file type; a PDF/A file is an ordinary PDF that follows extra rules and declares in its metadata which part and level it conforms to. Any PDF reader can open it.

The guiding idea is that a PDF/A file must be self-contained and unambiguous. Everything needed to display it identically, including fonts and color definitions, has to be inside the file, and nothing may depend on a particular program, server or setting that might not exist in the future.

What the standard requires and forbids

The exact list differs a little between parts, but the core rules are consistent:

• All fonts must be embedded. A document that relies on a font installed on the author’s computer may look different elsewhere, so this is not allowed.
• No encryption. A password could be lost, and an archive must stay readable for anyone authorized to see it. Password protection and permission restrictions are both forbidden.
• No JavaScript or executable actions. Scripts can change what the document shows and may not run the same way in future software.
• No external dependencies. Content may not be pulled in from outside the file, such as linked fonts or media streamed from a server. Audio and video are excluded.
• Device-independent color. The file must describe its colors precisely, typically by including an output intent with an ICC color profile.
• Standard metadata. Document information has to be stored in XMP metadata, which also carries the PDF/A identification.

Ordinary hyperlinks to websites are allowed, since they are just references and the page still displays correctly without following them.

Parts and conformance levels

PDF/A is published in several parts, each based on a different version of PDF, and each part has conformance levels marked by a letter.

The parts

• PDF/A-1 is based on PDF 1.4. It is the strictest: no transparency, no layers, and the older LZW compression is forbidden. Its strength is that almost every system that accepts PDF/A accepts part 1.
• PDF/A-2 is based on PDF 1.7 (ISO 32000-1). It allows transparency, optional content layers and JPEG 2000 compression, and permits embedding other PDF/A files. For most new documents it is the sensible default.
• PDF/A-3 is identical to part 2 except that it allows any file type to be embedded, such as a spreadsheet, a CAD file or XML data. Some e-invoicing formats use this to carry machine-readable invoice data inside a human-readable PDF.

A newer part based on PDF 2.0 also exists, but parts 1 to 3 remain the ones most commonly requested.

The levels

• b (basic) guarantees that the document will look the same in the future. It is the level most often asked for.
• u (Unicode) adds that every piece of text must map to Unicode, so it can be reliably searched and copied. It exists for parts 2 and 3.
• a (accessible) additionally requires a tagged logical structure, reading order and alternative text, so assistive technology can make sense of the document. Producing level a properly usually has to start in the authoring software, because structure cannot be reliably guessed afterward.

Who asks for PDF/A

PDF/A shows up wherever documents have to outlive the software that created them. Typical places include public archives and records offices, courts and electronic filing systems, government agencies, universities collecting theses, and industries with long retention duties such as construction, engineering, healthcare and finance. Many national archives list PDF/A among their preferred formats for long-term preservation.

Requirements vary a great deal, though, and they change. One system may demand PDF/A-1b specifically, another may accept any part, and some e-filing portals accept ordinary PDF as long as fonts are embedded and the file is not encrypted. In Turkey, if you are submitting through a system such as UYAP or an institution’s archive portal, check that system’s current technical guidance for accepted formats rather than assuming. When a recipient names a specific part and level, use exactly that.

Converting to PDF/A with vupdf

The vupdf PDF to PDF/A tool lets you choose PDF/A-1B, PDF/A-2B (recommended), PDF/A-2U, PDF/A-3B or PDF/A-3U. During conversion it embeds fonts that were not embedded when a suitable font is available, removes document-level JavaScript, adds the required color profile and metadata, and adjusts annotations and form fields to the PDF/A rules. A password-protected file can be converted after you enter its password; the result is unencrypted, as the standard requires.

Some content simply conflicts with a level. Attachments are only allowed in part 3, transparency is not allowed in part 1, and characters without a Unicode mapping rule out the u levels. By default the conversion stops and tells you why. If you turn on Allow lowering of the PDF/A conformance level, the tool instead picks the nearest level that fits the content and tells you which one it used.

After conversion you see a short report: what was repaired, which fonts could not be embedded, and any problem that might still make a strict validator reject the file. Read it; it tells you whether the result is likely to pass.

How to check conformance

A PDF/A label in the metadata is a claim, not a proof. Any software can write it. To be confident, check the file:

1. Run it through a validator. veraPDF is a free, open-source validator developed with the support of the archiving community and is widely used as a reference. It reports each rule that fails.
2. Open the file in a desktop reader. Many readers show a notice that the document conforms to PDF/A and is opened in a read-only mode.
3. Check the visual result page by page against the original, paying attention to special characters, form fields and anything that used transparency.
4. If the recipient has its own checker, as some filing systems do, use it before the deadline rather than on the last day.

Summing up

PDF/A trades flexibility for durability: embedded fonts, defined colors, no encryption, no scripts and no outside dependencies. Choose part 2b unless someone tells you otherwise, part 3 when you must keep attachments, and part 1b when a recipient insists on maximum compatibility. Then validate the result, because a document meant to last decades deserves a two-minute check today.