PDF vs PDF/A: Which One Do You Actually Need for Legal Archiving? (2026)
PDF/A is the version of PDF designed to still open correctly in 30 years. Here's when courts and regulators require it, what it actually changes under the hood, and how to convert.
Quick Answer
PDF/Ais the ISO-standardised version of PDF (ISO 19005) built for long-term archiving — every font is embedded, no JavaScript, no external dependencies, no encryption. You need it when filing in court (PACER, EU e-Justice, UK courts), submitting to regulators (FDA, ISO audits), or archiving any record that must still open correctly decades from now. For day-to-day use, regular PDF is fine. Convert with Word ("Save As" → ISO 19005-1), LibreOffice ("Archive PDF/A-2b"), or a server-side converter.
The 30-second version
A regular PDF is a snapshot of a document at the moment it was made. PDF/A is a snapshot with a promise: every ingredient needed to render the document correctly is baked into the file itself. No looking up fonts from your operating system. No reaching out to an external server. No JavaScript that depends on a 2008 browser plugin to behave. Open the same PDF/A in 2050 and it should render identically to today.
That promise is exactly what courts, regulators, and archives need. A judge in 2045 referencing a 2026 filing cannot afford for the document to render differently because Calibri changed its metrics or because a custom Acrobat plug-in stopped working. PDF/A removes the variables.
What PDF/A actually changes under the hood
Compared to a regular PDF, a PDF/A file must:
- Embed every font it uses.Including standard fonts like Helvetica that most readers have built-in. The reasoning: you don't know what fonts a 2050 reader will ship with.
- Define all colour in a self-contained way. Either embed an ICC colour profile, or use a standard colour space whose definition is part of the PDF/A spec. No referencing an external monitor profile or printer profile.
- Contain no encryption.An encrypted file with a forgotten password is unreadable, so PDF/A forbids it. If you need access control on an archived document, that's a separate layer (file-system permissions, repository ACL) — the PDF itself stays open.
- Contain no JavaScript, no audio, no video, no executable attachments. Anything that depends on a runtime environment that might not exist later.
- Use only allowed compression schemes. Flate (zlib) and CCITT and a few others — nothing proprietary or patented in a way that might prevent future implementations.
- Be tagged for accessibilityif it's the "accessible" flavour (PDF/A-2u, PDF/A-3u). This lets screen readers and future indexers extract the logical reading order.
None of these changes affect how the document looks. They affect whether the document is guaranteed to look the same in 30 years.
When you actually need PDF/A
Required by policy or law:
- US federal courts (PACER, CM/ECF). Most circuits accept regular PDF but specifically allow PDF/A; some chambers require it.
- European Court of Justice and EU member-state e-Justice portals. PDF/A-1 or PDF/A-2 mandated for most filings since 2010s.
- UK courts (HMCTS). PDF/A preferred for case bundles and trial documents.
- FDA submissions (USA). PDF/A required for certain regulatory dossiers; CDER/CBER specify PDF/A-1.
- ISO audit records, SOC 2, HIPAA evidence. Auditors expect long-lived, immutable formats.
- National archives (US NARA, UK National Archives, Library of Congress, Bundesarchiv). PDF/A is the de-facto preservation format.
- German tax records (GoBD). PDF/A-3 with embedded XML for digital invoices (ZUGFeRD, Factur-X) is mandatory for many B2B invoices in 2026.
Strongly recommended, not strictly required:
- Executed contracts you want to be confident still open correctly in 15 years
- Diplomas, transcripts, professional certifications
- Property deeds and title documents
- Medical records intended for lifetime retention
- Engineering drawings for safety-critical infrastructure
Probably overkill:a contract you're emailing today and a counterparty will counter-sign within a week; an invoice to a friend; an internal team report. Regular PDF is fine.
PDF/A flavours: 1, 2, 3 — and a / b / u
The version number (1, 2, 3) tells you which PDF spec PDF/A is based on:
- PDF/A-1 (ISO 19005-1, 2005) — based on PDF 1.4. The strictest. No transparency, no layers, no JPEG 2000. Maximum compatibility with old archive systems.
- PDF/A-2 (ISO 19005-2, 2011) — based on PDF 1.7. Adds transparency, layers, JPEG 2000, file embedding (one PDF/A-1 inside another PDF/A-2). The right choice for most new archives in 2026.
- PDF/A-3 (ISO 19005-3, 2012) — same base as PDF/A-2 but allows arbitrary file attachments. Used for hybrid invoices like ZUGFeRD / Factur-X where a structured XML payload travels inside the human-readable PDF.
The letter suffix tells you which conformance level:
- b — "basic". The document is visually reproducible. Text may be image-only (a pure scan); the file isn't tagged for accessibility.
- u — "Unicode". Every glyph maps to a Unicode character, so the text is searchable and copyable.
- a — "accessible". Tagged for screen readers, with explicit reading order. Highest bar; required for accessibility-mandated archives.
For most legal and regulatory use, PDF/A-2b or PDF/A-2uis the right pick. PDF/A-1b stays useful only when you're submitting to a system that explicitly requires the older version.
How to convert a PDF to PDF/A (free options)
If you're creating the document fresh in Word
- File → Save As
- Change format to PDF
- Click Options
- Tick ISO 19005-1 compliant (PDF/A)
- Save
Word produces PDF/A-1a output by default — sufficient for most court filings. If you need PDF/A-2 or PDF/A-3, export to regular PDF from Word and convert in step 2.
If you're using LibreOffice (Mac, Linux, Windows free)
- File → Export As → Export as PDF
- In the General tab, tick Archive (PDF/A)
- Pick PDF/A-2b from the version dropdown (recommended default)
- Export
LibreOffice's PDF/A export is built on the same engine that drives most enterprise PDF servers. The result passes veraPDF validation cleanly for almost all common document shapes.
If you have an existing regular PDF
Several command-line tools convert in place: qpdf --object-streams=disable input.pdf output.pdf plus font embedding via Ghostscript, or gs -dPDFA=2 -sDEVICE=pdfwrite -sPDFACompatibilityPolicy=1 for one-shot. For non-developers, our PDF compressor repacks files through the qpdf engine which gets you partway (object-stream cleanup, font subsetting); a dedicated PDF/A export from Word or LibreOffice is the most reliable cross-platform path.
How to verify a PDF is valid PDF/A
Two free options:
- Adobe Acrobatshows a blue PDF/A banner at the top of the document if the file declares PDF/A conformance. (It doesn't actually validate — just trusts the metadata flag.)
- veraPDF is the open-source reference validator. Drop your file in, get a PASS/FAIL report with detailed conformance violations. Free, runs on every OS, what most enterprise archive systems use behind the scenes.
If a file claims PDF/A but fails veraPDF, the archive will probably reject it. Validate before submitting to any regulator or court.
PDF/A archive checklist
- Pick the right version: PDF/A-2b for most legal archives, PDF/A-3 if you need to attach structured XML
- Embed all fonts (Word + LibreOffice do this by default when PDF/A is enabled)
- Remove encryption — apply access control at the storage layer, not in the PDF
- Strip JavaScript and any interactive form actions
- Run veraPDF to confirm conformance before submission
- Strip metadata only if your retention policy requires it — PDF/A allows metadata and many archives prefer to keep authorship info
- Keep an unconverted master copy in case you ever need to re-export for a different PDF/A flavour
Frequently asked questions
What is PDF/A?
An ISO-standardised version of PDF (ISO 19005) designed for long-term archiving. Every font is embedded, no JavaScript or external dependencies, no encryption, all colour information self-contained. A PDF/A file is always a PDF; a PDF file is not always a PDF/A.
Do I actually need PDF/A?
Required for court filings (PACER, EU e-Justice, UK courts), regulated industries (FDA, ISO audits), and long-term legal or financial archives. Recommended for any document you want to be confident opens correctly in 20+ years. For day-to-day sharing, regular PDF is fine.
PDF/A-1 vs PDF/A-2 vs PDF/A-3 — which one?
PDF/A-1 is the strictest and oldest (good for legacy archive systems). PDF/A-2 is the modern default for most archives — based on PDF 1.7, supports transparency and layers. PDF/A-3 allows file attachments (used for e-invoices like ZUGFeRD). For most new legal archives in 2026: PDF/A-2b.
How do I convert a regular PDF to PDF/A?
Word: Save As → PDF → Options → tick ISO 19005-1. LibreOffice: Export As PDF → tick Archive (PDF/A-2b). For existing PDFs, qpdf or Ghostscript on the command line, or re-export the source document through Word/LibreOffice.
Need a regular PDF first?
Convert from any source format — Word, Excel, images, HTML — using our free in-browser tools, then export to PDF/A from Word or LibreOffice. No subscription.
Browse PDF Tools →Why ImageAndPDF
100% Free
No hidden costs, no credit card, no signup required.
Private & Secure
Many tools process files in your browser; some features use secure server processing.
Instant Results
Cloud-powered processing. Most files done in seconds.
Works Everywhere
Any browser, any device. Nothing to install.
Ready to work with your files?
30+ free tools for PDFs, images, and documents. No signup needed.
Browse All Tools
Kummari Achyuth is a software engineer and the founder of ImageAndPDF. He started the project after running into the same frustration most people meet with online file tools — uploads to unknown servers, daily limits, watermarks, and signups before any work could be done. His response was to build a suite of utilities that run almost entirely in the browser, using open-source libraries like pdf-lib, PDF.js, and ONNX Runtime, so files never have to leave the device for most operations. He works primarily on the platform's performance and privacy architecture: the rendering pipeline, the in-browser processing pathways, and the on-device AI models for background removal and image upscaling.