File formats

What PDF/A Is And When You Actually Need It

A scanned page saved as PDF/A is a conforming file containing a photograph. It will look identical in fifty years and nobody will be able to search it then either.

7 min read

PDF/A is the archival version of PDF, defined by what it refuses to allow. PDF/A is ordinary PDF with things forbidden. Everything a page needs must be inside the file. Anything depending on the outside world is banned. That single rule is the whole standard. It exists so a document opened in forty years looks exactly as it does now.

It also promises less than people assume. Conformance is about rendering and about nothing else at all.

PDF/A Is a Standard Made of Refusals

Most standards add capability. This one removes it. The list of what it takes away explains what it is for.

Must Be Inside

  • Every font the document uses
  • A definition of what the colours mean
  • Any image the page shows
  • Metadata in a standard form

Not Allowed

  • Fonts merely referenced by name
  • JavaScript and launched actions
  • Audio, video and embedded executables
  • Encryption
  • Anything fetched over a network
Read the right column as a list of dependencies. Each one is something that could be missing or changed or simply unavailable later. Every one is a reason files render differently. Or fail to render.

Encryption being forbidden surprises people. Lose the key and everything is gone, which is exactly the failure the standard exists to prevent. Protection belongs around the archive rather than inside.

Why Fonts Are the Heart of PDF/A

The font rule does most of the work and it is worth understanding why.

An ordinary PDF can name a font rather than carry it. Pages saying use Helvetica meet a machine that displaying it supplies one. Where that machine has a different Helvetica, it substitutes. Where it has nothing resembling one, it substitutes anyway and the layout shifts.

font referenced by name
  machine A  the intended font, correct layout
  machine B  a substitute, lines wrap differently
  machine C  no match, boxes or a default face

font embedded in the file
  every machine  identical, because the shapes
                 travelled with the document

This is the same distinction as linked and embedded images in mail. Something referenced depends on a resource that may not be there. Something carried does not. The pattern turns up wherever files have to survive being moved. The answer is always the same.

What PDF/A Conformance Does Not Promise

This is where organisations get caught, usually years later.

Conformance meansIt does not mean
The page will render the sameThe text can be searched
Nothing external is neededThe content is accurate
The file is self containedAnything was checked by a person
Forbidden features are absentUseful features are present
It meets a technical ruleIt meets your retention obligation

Conformance and content are different questions. A scan saved this way is a valid file holding a photograph. It will look identical in fifty years. Nobody will be able to search it then either. An archive of ten thousand of those satisfies a conformance check and answers no question anybody asks of an archive.

Where the PDF/A Conformance Levels Come In

The parts of the standard track the version of PDF underneath. Levels within each describe how much a file records about its own structure.

  1. The basic level Renders reliablySelf contained. Identical wherever it opens. Nothing recorded about what is a heading or what order to read in
  2. The accessible level It will look right and be readableAdds tagging, so the structure and reading order are recorded. What a screen reader needs. Also what makes extraction work
The higher level is the one worth asking for. Tagging separates files that survive from files that can still be used. It costs nothing extra at the moment of creation.

Which level applies is usually somebody else's decision. A regulator, a court or an archive normally specifies it. Where nothing has been specified and the choice is yours, the accessible level is the one that will not need doing again.

How PDF/A Fits the Longevity Question

The site has already set out what makes a format survive. Public specification, text where possible, several independent readers. PDF/A is an interesting case against those tests.

TestPDF/A
Published specificationYes, an ISO standard
Text or binaryBinary, which counts against it
Independent readersMany. It is ordinary PDF underneath
Depends on anything externalNo, by design

It fails one test and answers another that the original three do not cover. A plain text file is more durable and tells you nothing about how it should look. PDF/A is less durable in principle and preserves an appearance, which is sometimes the thing that matters.

The choice is about what you are preserving. Where the content is the point, keep it in the most open format that holds it. Where how it looked is the point, as with a signed contract or a stamped certificate, that is the case PDF/A was built for.

Read next Which Mail Formats Last And Which Already Died The three tests this standard is being measured against.

Deciding Whether You Need PDF/A

  1. Ask whether anybody specified it. Regulators, courts and archives often do. Where somebody has, that answers the question and the level as well.
  2. Ask how it will be used later. Read once and discarded? Or produced as evidence in a decade? Only the second justifies the cost.
  3. Check whether there is text in it at all. Converting a scan to PDF/A preserves a picture. Run recognition first if the words need to be findable.
  4. Ask for the accessible level where the choice is yours. Tagging costs nothing at creation and cannot be added convincingly afterwards.
  5. Validate rather than assume. Software saying PDF/A and a file passing a conformance check are different claims. Only one of them has been tested.

The commonest mistake is converting the wrong thing. Organisations put considerable effort into making scanned archives conform, producing a large collection of guaranteed pictures. The conversion was done correctly. It was done to files that had nothing worth preserving beyond the fact that they existed.

What the Standard Is Really Saying

Underneath the rules there is a single idea. It applies well beyond documents.

Files that depend on something else will eventually meet a moment when that something is absent. A font that was installed everywhere in 2005, a colour profile, a server that answered a request. None of them looked like a risk at the time. Each is a way for a document to stop being what it was.

PDF/A is a list of things somebody noticed could go missing. That is a more useful way to read it than as a compliance box. It also explains why the rules feel arbitrary until you see what each is defending against.

Conformance rules checked against the referenced articles in August 2026. Requirements differ between the parts and levels of the standard, so treat this as the shape of it rather than a conformance checklist.

Questions People Ask

7 questions, answered in full below.

What is PDF/A?

An ISO standard describing a version of PDF with things taken away. Everything a page needs sits inside the file. Anything depending on the outside world is forbidden, so rendering stays the same in decades as it does now.

How is it different from an ordinary PDF?

By what it refuses. Fonts travel with it rather than being named. Colour has to be defined rather than assumed. No JavaScript, no audio or video, no encryption, nothing fetched from a network. What comes out relies on nothing outside itself.

Does PDF/A mean the text is searchable?

No. This is the commonest misunderstanding of the standard. A scanned page saved as PDF/A is a conforming file containing a picture. The standard promises the page will look the same. It promises nothing about whether there is any text in it.

Do I need it for ordinary documents?

Only where something has to be readable long after everybody involved has moved on. Or where a regulator or an archive has asked for it. For a document somebody will read this month it adds size and takes away features for no benefit.

Why can a PDF/A file be so much larger?

Because every font it uses travels with it rather than being borrowed from the machine displaying it. That is the point rather than a fault. It separates files that render anywhere from ones that renders where the fonts happen to be installed.

Can I convert an existing PDF to PDF/A?

Usually. It is a real conversion rather than a relabelling. Fonts get embedded, colour gets defined, forbidden features are removed. Anything that cannot be made to conform, such as a font that was only referenced, has to be substituted or the conversion fails.

What are the different parts and levels?

The parts follow whichever version of PDF sits beneath them. Levels say how much structure a file records. Higher levels add tagging, so reading order and headings are written down. Which one applies is normally something the receiving side specifies.

Sources

Where the figures and behaviour described above were checked.

  1. PDF/A Wikipedia
  2. Digital preservation Wikipedia
  3. Font embedding Wikipedia