Convert PDF to PDF/A: Long-Term Archival Format Explained (2026 Guide)

A grant officer emails you: “Please re-submit the final report as PDF/A-2b. Our archive doesn’t accept regular PDFs.” You stare at the email. You’ve used PDF for twenty years. What is PDF/A, why is there a “/A”, and why does the version number have a letter on the end of it?

This is the most common first encounter with PDF/A — in a compliance email from a government agency, a court clerk, a research funder, or an institutional archive. The format isn’t difficult, but the standard is genuinely picky about what it allows, and most “convert to PDF/A” tools don’t tell you what they’re stripping out.

This guide walks through what PDF/A actually is, what each level (1a, 1b, 2b, 2u, 3b, 4) means in practice, what your file loses on the way in, and when you genuinely need it vs when a regular PDF is fine.

What PDF/A is

PDF/A is the ISO 19005 standard, first published in 2005, for the long-term archival of electronic documents. It is a restricted subset of regular PDF — every PDF/A file is a valid PDF, but not every PDF is a valid PDF/A.

The restriction has one design goal: a PDF/A file must render the same way in 50 years as it does today, on software that doesn’t exist yet, without needing any external resource.

Think of it as PDF with a strict diet. A regular PDF can:

• Reference fonts that are installed on the reader’s computer
• Run JavaScript when opened
• Embed video and audio
• Link to external files
• Be encrypted

A PDF/A file cannot do any of those except embedding files (and even that only in PDF/A-3+). Everything the document needs to display correctly must be inside the file itself: fonts embedded, colours defined with explicit colour profiles, no scripts, no external dependencies.

The trade-off: a slightly larger file in exchange for a file you can trust to open and render identically in 2076.

Why governments, libraries, and courts require it

The reason boils down to a simple worry: software changes, fonts disappear, web links break, plugins are deprecated. A regular PDF made in 2010 referencing a corporate font and a Flash animation is, in 2026, a partially broken document. A regular PDF made in 2026 referencing a modern font and a piece of JavaScript may be similarly fragile in 2046.

For a document that has to be readable decades later — a court filing, a medical record, a tax return, a published research dataset, a parliamentary report, a UNESCO archive item — that’s not acceptable. PDF/A removes the external dependencies and locks down the rendering, so the file is verifiably self-contained.

In practice, you’ll meet PDF/A requirements from:

• Court e-filing systems — most US federal and state courts, EU national courts, and the UK Online Court
• National and academic archives — Library of Congress, British Library, EU national archives, university institutional repositories
• Government grant reporting — many EU Horizon programmes, US federal grants, Australian and Canadian research councils
• Records retention under regulation — SOX records, GDPR-relevant document retention, HIPAA medical records, ISO 19005-compliant document management systems
• Patent offices — USPTO, EPO, WIPO for patent submissions and prior-art archives

If you’re submitting to any of these and not sure of the level, PDF/A-2b is almost always accepted.

The levels — without drowning you

The version numbers look intimidating. Once you know the pattern, they’re simple.

Format: PDF/A-[version]-[conformance level].

Version = which PDF/A revision (1, 2, 3, 4):

• PDF/A-1 (ISO 19005-1, 2005) — based on PDF 1.4. Most restrictive. Used for legacy long-term archives.
• PDF/A-2 (ISO 19005-2, 2011) — based on PDF 1.7. Adds JPEG 2000, transparency, OpenType fonts, optional content (layers). The modern default.
• PDF/A-3 (ISO 19005-3, 2012) — same as PDF/A-2 plus permits embedded attachments of any type. Used for invoice formats like ZUGFeRD/Factur-X that bundle XML inside PDF.
• PDF/A-4 (ISO 19005-4, 2020) — based on PDF 2.0. Modernised, drops the level letters in favour of named flavours (PDF/A-4, PDF/A-4e for engineering, PDF/A-4f for embedded files).

Conformance level = how strict (a, b, u):

• b (basic) — guarantees visual reproduction. The file looks the same forever. Doesn’t require accessibility tagging or Unicode mapping.
• u (unicode) — basic + Unicode mapping. Every character in the file maps to a Unicode value, so text is searchable and copy-pasteable forever.
• a (accessible) — unicode + full accessibility tags (structure, reading order, alt text on images). Required for accessibility-mandated archives.

So:

• PDF/A-1b — visual reproduction, old PDF base, most conservative
• PDF/A-1a — 1b + accessibility
• PDF/A-2b — visual reproduction, modern PDF base, practical default in 2026
• PDF/A-2u — 2b + searchable text
• PDF/A-2a — 2u + accessibility tagging
• PDF/A-3b/u/a — same as 2 series but allows attachments
• PDF/A-4 — modern, PDF 2.0 base, for new long-term archives

If the receiving party didn’t specify, PDF/A-2b is the safe answer. If they need searchable text (most archives do), PDF/A-2u. If they have accessibility mandates (most public sector EU and US federal agencies), PDF/A-2a.

What gets stripped out

Converting a regular PDF to PDF/A is a lossy operation in a specific sense — anything not allowed by the spec gets removed. The list:

• JavaScript — any embedded scripts, including form validation, popups, calculated fields. Form fields stay; the scripts behind them do not.
• Audio and video — sound clips, embedded MP4s. Stripped (allowed only in PDF/A-4 with caveats).
• External file links — links to files on the user’s filesystem or on the network. Hyperlinks to web URLs are allowed because they’re text references, not embedded resources.
• Encryption — passwords, DRM, signed-but-encrypted files. The whole file becomes unencrypted. Long-term archives handle access control at the storage layer.
• Non-embedded fonts — fonts referenced from the system are replaced by embedded copies, or substituted with a free font with similar metrics if the original isn’t available.
• LZW compression (PDF/A-1 only) — replaced with Flate.
• Transparency and layers (PDF/A-1 only) — flattened. PDF/A-2 and later allow these.

What you keep: text, images, vectors, hyperlinks to web URLs, form fields (without scripts), digital signatures (with caveats — see below), and metadata.

The conversion also adds things: the PDF/A identifier in metadata (so verifiers can recognise the format), explicit colour profiles (sRGB by default), and embedded fonts where they were previously referenced.

When you genuinely need PDF/A

You need PDF/A when:

• A government, court, archive, or grant body explicitly asks for it
• You are running an internal long-term records system with a retention period over five years
• You’re publishing research data, government reports, or any document that has to be cite-able indefinitely
• You’re producing invoices in regulated formats (ZUGFeRD, Factur-X) that bundle structured XML — these are PDF/A-3 by spec
• Your document management system enforces ISO 19005 compliance on ingest

You don’t need PDF/A for:

• Day-to-day operational documents (meeting notes, project drafts, internal memos)
• Email attachments to colleagues
• Documents you’ll edit again soon
• Print-and-discard documents
• Anything where the receiver hasn’t asked for it

Converting everything to PDF/A “just in case” is wasteful — file sizes grow, JavaScript form helpers disappear, and you lose features you might still use. Reserve PDF/A for the documents that actually need long-term self-contained archival.

The step-by-step (in-browser, free, no signup)

1. Open PDF to PDF/A — runs entirely in your browser; the document never uploads
2. Drop your PDF in
3. Pick the PDF/A level the receiver requested. If they didn’t specify, pick PDF/A-2b.
4. The tool will warn you about anything that has to be stripped (JavaScript, attachments, encryption) before conversion — review and confirm
5. Click Convert to PDF/A and wait — for a normal document this is seconds
6. Download the resulting .pdf file (yes, the extension stays .pdf; PDF/A is identified by metadata, not by extension)
7. Verify with a PDF/A validator (free options: VeraPDF, the validator built into Acrobat) before submitting, especially the first time

That’s it. No upload, no signup, no watermark, no daily quota.

Verifying a PDF/A file

A PDF/A file claims compliance in its metadata, but the only way to be sure is to validate it. The free open-source VeraPDF validator is the de facto standard — it’s the same engine many archives use on ingest. Drop your file in, get a pass/fail with line-by-line details of any non-compliance.

Things that commonly fail:

• A font that didn’t embed cleanly
• A colour space without an explicit profile
• A form field with a remnant script
• An interactive element (button, action) that snuck through

Validation takes seconds and saves you the round-trip of submitting an “almost-compliant” file that gets rejected.

Common pitfalls — and how to fix them

Pitfall 1: Conversion fails because the source has fonts the tool can’t embed. Re-export the source from the original application with “embed all fonts” enabled, or substitute with a standard PDF font (Helvetica, Times, Courier) which always embeds. The visual result may shift slightly; that’s the cost of long-term reliability.

Pitfall 2: Form fields stop working after conversion. The validation scripts behind them were stripped. If the form needs to be fillable in the archive, the fields themselves remain interactive; only the calculated logic is gone. For a form that needs computed totals, fill it before converting to PDF/A.

Pitfall 3: Hyperlinks to web URLs work; links to local files don’t. That’s by spec. PDF/A allows web URLs (they’re text) and forbids local file links (they’re external dependencies). Replace local references with absolute web URLs where possible.

Pitfall 4: Encrypted PDF rejected. Decrypt first using your password, then convert. PDF/A is by definition not encrypted; storage-layer protection (encrypted disk, access control) is the right place for long-term security.

Pitfall 5: The receiver still rejects it. Run VeraPDF, share the validation report with the receiver, and ask which line specifically fails. Sometimes the receiver’s system requires a sub-flavour (e.g., specifically PDF/A-2u for searchability, not just 2b).

PDF/A and digital signatures

A digital signature applied to a PDF before PDF/A conversion may not survive — most converters re-write the document structure, which invalidates any cryptographic seal. The reliable order is:

1. Convert to PDF/A first
2. Then apply the digital signature
3. The signature itself must conform to PDF/A’s allowed signature types (PAdES, common in EU eIDAS flows)

For typed/drawn/image (electronic) signatures, the order matters less because they’re visual stamps without cryptography, but the convention is the same: sign last.

A quick comparison of free options in 2026

Tool	Where files go	PDF/A levels supported	Validator included	Watermark
imisspdf — PDF to PDF/A	In your browser	1b, 2b, 2u, 2a, 3b, 3u	Basic	None
ILovePDF (free tier)	Server upload	1b, 2b, 3b	No	None
Smallpdf (free tier)	Server upload	1a, 2b	No	After daily limit
Adobe Acrobat Online	Server upload	1a, 1b, 2a, 2b, 2u, 3a, 3b, 3u	Yes	None on free
VeraPDF (desktop)	Local	Validation only	Yes, gold-standard	None

The combination most organisations land on in 2026: a privacy-preserving in-browser tool for the conversion, plus VeraPDF on the validation side. Free, local, defensible.

A note on the long view

PDF/A asks you to give up a few PDF features (scripts, video, external links) for a file that will still open and look right in 2076. In 2026, that’s an abstract benefit — none of us will know until decades from now whether the bet paid off.

But that’s the point of an archival format: you make the conservative choice now so the future doesn’t have to. Court records, medical records, research datasets, government reports — these are the documents where the long view is the only view that matters.

For everything else, regular PDF is fine.

Frequently asked questions

The FAQ block at the top of this article covers the most common questions about PDF/A conversion. If you’re not sure which level you need, ask the receiving party — and if they don’t know either, PDF/A-2b is the safe modern default.

Try the tool

When you’re ready: Convert PDF to PDF/A →. Drop your PDF in, pick the level, download the compliant archive copy seconds later — no upload, no signup, no watermark.

Frequently asked questions

Related articles