PDF/A is an ISO-standardized subset of PDF (ISO 19005) designed for long-term archiving. A normal PDF can silently depend on fonts installed on your computer, network resources, JavaScript, or encryption — any of which can break in ten years. PDF/A forbids all of that and requires everything to be embedded in the file itself, so the document renders identically decades from now on any conforming reader. If a court, university, government office, or EU e-invoicing rule asks for "PDF/A", this guide covers exactly what they mean and how to comply for free.
Published 20 October 2026 · Updated 20 October 2026
Open a fifteen-year-old PDF and you might find missing fonts, broken embedded graphics, or a demand for a plugin that no longer exists. Ordinary PDFs are allowed to reference external resources — a font on the author's machine, a color profile on a server, content fetched at view time. When those resources vanish, the document degrades. PDF/A's insight is simple: for archival copies, the restriction is the feature. A conforming PDF/A file is 100% self-contained — every font, color profile, and resource it needs is embedded inside it. No encryption, no JavaScript, no audio or video, no external references. Any conforming reader, decades from now, should render it exactly as it looks today.
Note what PDF/A does not promise: it is a preservation container, not a records-management policy. A valid PDF/A file can still contain the wrong document, weak metadata, an incomplete scan, or unscannable text. Retention rules, checksums, backups, and integrity checks remain your separate responsibility.
| Version | Standard | Key differences |
|---|---|---|
| PDF/A-1 | ISO 19005-1 (2005) | Oldest and strictest. No transparency, no layers. Two conformance levels: A (accessible, full) and B (basic, visual only). |
| PDF/A-2 | ISO 19005-2 (2011) | Allows transparency, layers, JPEG 2000, and embedding of other PDF/A files. The common default for business documents — PDF/A-2b is the safest choice when nobody specified otherwise. |
| PDF/A-3 | ISO 19005-3 (2012) | Same as A-2, but allows embedding any file type as attachments. This is how EU e-invoicing formats ZUGFeRD and Factur-X carry machine-readable XML inside a human-readable invoice — the reason many German companies must use A-3. |
| PDF/A-4 | ISO 19005-4 (2020) | Modernized on the current PDF specification (ISO 32000-2). Adds conformance levels 4e (embedded files) and 4f. Slowly replacing A-2 in new procurement specs. |
The practical rule: if a client says "we need PDF/A," the follow-up question is always "which level." For EU e-invoicing it is usually A-3 (because of the embedded XML). For legal and government archives, A-2b remains the most requested. Ask before you convert — the wrong conformance level fails acceptance tests just as surely as no PDF/A at all.
Courts and discovery protocols in many jurisdictions require PDF/A for produced documents. Self-contained files guarantee that the record the court sees in 2036 is the record you produced today — pair with Bates numbering for the full discovery workflow.
Archives, land registries, and public-sector retention systems increasingly specify PDF/A because retention periods run 10–30+ years — longer than any software license.
German and French e-invoicing standards embed machine-readable XML inside a PDF/A-3 invoice. If you invoice German clients electronically, A-3 is not optional — it is the format.
University repositories and journal archives ask for PDF/A so dissertations and papers stay readable for the institution's lifetime. Font embedding matters most here — theses use unusual symbols and fonts.
If nobody asked for PDF/A, do not use it as your everyday format: the restrictions conflict with active workflows (encryption, collaboration, form filling). Archive in PDF/A, work in regular PDF.
Archives are forever, and that is exactly why the pre-archive cleanup matters. Ordinary PDFs carry hidden passengers: author names, editing software, revision history, comments, and sometimes entire deleted passages left in the file structure. Once a document enters a PDF/A archive — a court record, a university repository, a government system — that hidden data becomes part of the permanent record. Strip it before it is immortalized: use our Remove PDF Metadata tool to see and delete hidden author/title/software data, and Redact PDF to permanently burn out privileged text before archiving.
The privacy contrast with the big platforms is worth stating plainly. Adobe's cloud tools, Smallpdf, and iLovePDF all receive your full document on their servers before processing it — for archival workflows that means your most permanent, most sensitive records pass through a third party's infrastructure and its retention policies. A browser-only tool processes the file on your own device: there is no server copy, no retention window, no account tying your identity to the document. For the documents you keep forever, "never uploaded" is the only privacy posture that actually matches the stakes.
PDF/A stands for Portable Document Format / Archival. It is an ISO-standardized subset of PDF (ISO 19005) designed for long-term document preservation — a PDF/A file is guaranteed to render the same decades from now because everything it needs is embedded inside it.
A normal PDF may depend on external fonts, encryption, JavaScript, audio/video, or files fetched over the network — any of which can break over time. PDF/A forbids all of these and requires every font, color profile, and resource to be embedded in the file, making it completely self-contained and future-proof for archiving.
PDF/A-1 (2005) is the oldest and strictest. PDF/A-2 (2011) allows transparency and layers and is the common default — PDF/A-2b is the recommended choice when nobody specified a level. PDF/A-3 (2012) additionally permits embedding arbitrary files, which is how EU e-invoicing formats like ZUGFeRD/Factur-X carry machine-readable XML inside the invoice. PDF/A-4 (2020) modernizes the standard.
When a court, regulator, archive, employer, or client specifies it. Common cases: legal filings and case files, government records, EU e-invoices (ZUGFeRD/Factur-X require PDF/A-3), thesis submissions to university repositories, and long-term business archives. If nobody asked for PDF/A, a regular PDF is usually the better working format.
Microsoft Word, Excel, and PowerPoint all have a free "Save as PDF/A" (ISO 19005-1 compliant) option in their PDF export dialog — tick "ISO 19005-1 compliant (PDF/A)". LibreOffice offers PDF/A export too. For existing PDFs, Adobe Acrobat Pro can convert with a preflight check. After converting, validate the result with a free checker like veraPDF to be sure it conforms.
Password protection is forbidden in PDF/A — archives must stay accessible, so encryption makes a file non-conformant. Editing a PDF/A file breaks its conformance too. The archival workflow is: finish the document, convert to PDF/A, then store it read-only.
Yes — image-based scans archive well in PDF/A. Two tips: deskew crooked scans before converting, and run OCR first if you want the archived copy to be searchable, since PDF/A itself does not add searchable text.