PDF/A explained

Archival, not print. Self-contained files built to outlive their software.

ISO reference	ISO 19005-1:2005, 19005-2:2011, 19005-3:2012
Based on	PDF 1.4 for A-1, PDF 1.7 for A-2 and A-3
Purpose	Long-term archiving
Transparency	Not allowed in A-1, permitted from A-2
Colour spaces	Device-independent, colour must be unambiguously defined

PDF/A solves a different problem from PDF/X. PDF/X asks will this print correctly at that shop. PDF/A asks will this still open and look identical in thirty years, on software nobody has written yet.

The answer, in both cases, is achieved by removing external dependencies.

The core rules

Everything embedded. Fonts, colour profiles, images. Nothing may be fetched or assumed.
No encryption. A file nobody can open is not archived.
No JavaScript, no audio or video, no executable content.
Colour unambiguously defined, so device-dependent values cannot be misinterpreted later.

The three generations

PDF/A-1 (2005, based on PDF 1.4) is the strictest. No transparency, no layers, no JPEG 2000. Files conforming to it are extremely conservative and extremely portable.

PDF/A-2 (2011, based on PDF 1.7) permits transparency, layers and JPEG 2000, and allows embedding other PDF/A files. For most new archival work this is the sensible target.

PDF/A-3 (2012) adds one significant capability: it may embed arbitrary file types, not just PDF/A. That is what makes electronic invoicing formats such as Factur-X and ZUGFeRD possible, where a human-readable invoice carries the machine-readable XML inside it.

The conformance levels: a, b and u

This is the part that confuses people, and it is simpler than it looks.

b (basic) guarantees visual reproduction. The file will look the same in future. It says nothing about structure.
a (accessible) additionally requires tagged structure, a defined reading order and Unicode mappings, so the content is machine-readable and usable with assistive technology.
u (from A-2 onward) sits between them: Unicode mappings required, but not full tagging.

PDF/A-1b is by far the most commonly requested, because it is achievable from almost any source document. PDF/A-1a requires the source to have been authored with real structure, which most are not.

When it applies to you

Usually because somebody requires it: court e-filing, government records, regulated industries and long-term document retention. If a submission portal rejected your file, the requirement is almost always a specific PDF/A level, and the rejection message will name it.

Check a file against PDF/A

Conformance is not something you can confirm by eye. Upload a PDF and Filecheck reports which standard it actually conforms to, and what would have to change. Free, no signup, files deleted after processing.

Validate a PDF Full preflight check

Related concepts

Embedded Fonts Why embedding font programs is the single most important factor for text rendering reliability in print. PDF Security & Signatures Why password locks, restricted permissions, and digital signatures block prepress automation, and how Filecheck handles them. PDF Transparency Why complex transparency effects, drop shadows, and opacity settings require flattening for reliable printing.

The other standards

PDF/X-1a The strictest and oldest. CMYK only, everything flattened. PDF/X-3 Like X-1a, but colour-managed RGB is permitted. PDF/X-4 The modern default. Live transparency, no destructive flattening. PDF/UA Accessibility. Tagged structure, reading order, alternative text.