A normal PDF can break over time — fonts go missing, links rot, passwords get lost. PDF/A is the archival version of PDF: self-contained, font-embedded and built to render identically decades from now. This guide shows the free ways to convert, which version to pick, and how to verify the result.
Quick answer: the free way is LibreOffice — open the document, choose File → Export As → Export as PDF, tick "PDF/A-2b", export, then validate the file with veraPDF. Microsoft Word can do the same from its PDF export options ("ISO 19005-1 compliant (PDF/A)"). No browser tool — including ours — can produce a fully compliant PDF/A file, so use a desktop tool and verify the result.

PDF/A is the ISO 19005 family of standards for long-term preservation of electronic documents. A PDF/A file is a normal PDF with strict rules added: every font used must be embedded, colours must be device-independent (an output intent or ICC profile), metadata must be present in XMP format, and encryption, JavaScript, audio, video and external references are all forbidden. The result is a file that contains everything it needs to render — nothing is borrowed from the reader's computer, so the document looks the same in fifty years as it does today.
The need is real. A PDF that references a font installed on your computer today renders differently — or breaks — on a machine without that font. A PDF with an embedded video stops working when the video format dies. An encrypted PDF becomes unreadable the day the password is forgotten. Archives, courts, governments and businesses learned this the hard way, which is why PDF/A exists: it trades features for permanence. As the veraPDF Consortium puts it, veraPDF is "the open-source PDF/A and PDF/UA validator" — the reference check that archives and digital preservation institutions rely on to prove a file really is what it claims to be.
PDF/A has four parts, published between 2005 and 2020, each built on a newer PDF version:
| Part | Standard | Based on | What it adds |
|---|---|---|---|
| PDF/A-1 | ISO 19005-1 (2005) | PDF 1.4 | The original. No transparency, no layers, no JPEG 2000, no embedded files. Best for simple text and scans. |
| PDF/A-2 | ISO 19005-2 (2011) | PDF 1.7 | Adds transparency, JPEG 2000, layers and embedding of other PDF/A files. Level 2b is the most widely adopted standard today. |
| PDF/A-3 | ISO 19005-3 (2012) | PDF 1.7 | As PDF/A-2, plus embedded files of any type — the basis of Factur-X and ZUGFeRD e-invoices, where an XML invoice travels inside the PDF. |
| PDF/A-4 | ISO 19005-4 (2020) | PDF 2.0 | Built on PDF 2.0. Drops the a/b/u levels for plain 4, 4e (engineering, 3D) and 4f (embedded files). |
Within parts 1–3, the conformance levels matter too. Level b (basic) requires visual reproducibility — the file looks the same forever. Level u adds a Unicode mapping for all text, so the file stays searchable forever. Level a (accessible) adds full structural tagging, so the file stays accessible forever. PDF/A-4 reorganises this into plain 4, 4e and 4f, and accessibility is expressed by also claiming PDF/UA.
For most archiving, the answer is PDF/A-2b. It supports transparency and modern compression, it is the most widely adopted standard today, and every serious validator and archive accepts it. Pick PDF/A-1b only when the receiving archive explicitly demands it, or for simple scanned documents where the extra features buy nothing. Pick PDF/A-3 when the source data must travel inside the same file — an XML invoice, a spreadsheet, the original Word document. Pick PDF/A-4 when your workflow already uses PDF 2.0 features or engineering content such as 3D models. And pick level u over b whenever you can: searchable text costs nothing and saves the archive from becoming a pile of pictures.
One warning that applies to every version: converting a file to PDF/A does not fix a bad source. A scan of a blurry photocopy becomes a blurry PDF/A. A document with missing fonts becomes a PDF/A with substituted fonts baked in. Convert from the cleanest original you have — the Word document or the high-resolution scan — not from a fifth-generation copy.
LibreOffice is the best free option for most people: it is a full office suite, it runs on every desktop system, and its PDF/A export embeds fonts and writes the XMP metadata for you. If you are starting from a Word document, open it in LibreOffice and export the same way — the result is identical.
In Word, choose File → Save As → PDF, then click Options and tick "ISO 19005-1 compliant (PDF/A)". Word produces PDF/A-1a — the oldest part, but perfectly valid for text documents, and it carries the accessibility tagging that level a requires. Use this when the document starts life in Word and you want the simplest possible path.
For converting many files at once, Ghostscript is the free workhorse. A basic conversion looks like this:
gs -dPDFA=2 -dBATCH -dNOPAUSE -sColorConversionStrategy=RGB \ -sDEVICE=pdfwrite -dPDFACompatibilityPolicy=1 \ -sOutputFile=archived.pdf input.pdf
Ghostscript is powerful and free, but it is a command-line tool: it will warn you — loudly — about fonts it cannot embed, which is exactly the information you need. If it reports unembedded fonts, the output is not compliant, and no amount of re-running will fix a font that isn't on your system. Install the missing font, then convert again.
PDF24 Tools, the German free PDF suite, offers PDF/A conversion on Windows without an upload. Adobe Acrobat Pro has the most complete PDF/A support of any desktop tool — preflight checks, conversion and validation in one place — but it is a paid subscription, and the free methods above cover the same ground for ordinary archiving. What none of these methods is: a website you upload to. If a "free PDF/A converter" site asks for your file, it is doing the conversion on its server, and your archive document has just left your control.
This is the part other sites won't tell you. A fully compliant PDF/A file needs three things a browser cannot reliably provide: fonts embedded from font files the browser doesn't have, an XMP metadata packet declaring the conformance level, and a colour output intent. Browser PDF libraries such as pdf-lib — the engine behind our own tools — can build excellent ordinary PDFs, but they cannot embed arbitrary fonts or write the XMP declaration that makes a file PDF/A. Any website that claims to convert your PDF to PDF/A in the browser is either producing a file that fails validation, or uploading your document to a server to do the work there.
That is why this guide has no tool widget, and why our site does not offer a PDF/A converter. The honest route is a desktop tool — LibreOffice, Word, Ghostscript or Acrobat — followed by validation. Our PDF/A explainer covers the standard in more depth, and our metadata editor and metadata remover handle the ordinary metadata jobs that browser tools do well.
Never trust a converter's own claim. Validate every archived file with veraPDF, the open-source reference validator maintained by the veraPDF Consortium and recommended by archives and digital preservation institutions. It is free, it runs as a desktop application and a command-line tool, and it checks the file against the ISO 19005 rules for PDF/A-1a through PDF/A-4f, reporting pass or fail with the exact rule that failed. A file that passes veraPDF is PDF/A; a file that doesn't is a normal PDF with a hopeful filename.
Validate at creation, on receipt, and before permanent storage — and recheck after any edit, since merging pages, adding annotations or applying a password can all break compliance. And remember that PDF/A does not protect against corruption or hardware failure: keep backups, checksums and a migration plan alongside your validation reports. Archiving is a process, not a file format.
Archival format and backup are different jobs. PDF/A keeps a file readable; it does not keep it safe from disk failure, fire or theft. Keep at least two copies in different places, with checksums, and re-validate on a schedule.
A PDF/A file cannot be password-protected — encryption is forbidden in every part of the standard. Sign or secure the working copy before you convert it for the archive, and never convert an encrypted file and expect it to pass.
Converting a blurry scan or a font-missing document produces a blurry, font-substituted PDF/A. Always convert from the cleanest original, and read the validator's report — it tells you exactly what failed and why.
PDF/A is the ISO 19005 family of standards for long-term archiving. A PDF/A file is a self-contained PDF: every font is embedded, colours are device-independent, and encryption, JavaScript, audio, video and external references are forbidden — so the file renders identically decades from now, on any device.
No. A normal PDF can reference fonts installed on the reader's computer, link to external content, and carry passwords and scripts. A PDF/A file may do none of those things — everything it needs to render is inside the file, which is what makes it safe to archive.
For most archiving, PDF/A-2b: it supports transparency and modern compression, and it is the most widely adopted standard today. Use PDF/A-1b only for simple scanned documents, PDF/A-3 when you must embed the source file (for example an XML invoice), and PDF/A-4 when your workflow already uses PDF 2.0.
Honestly, not reliably. True PDF/A needs embedded fonts, an XMP metadata declaration and a colour output intent — work that browser tools cannot do properly, including ours. Use a free desktop tool instead: LibreOffice exports PDF/A-2b directly, and Ghostscript converts from the command line. Both are free.
No. Encryption is forbidden in every PDF/A part, because an archive must never be locked behind a key that could be lost. Sign or secure the working copy before you convert it for the archive.
Validate it with veraPDF, the open-source reference validator maintained by the veraPDF Consortium. It checks the file against the ISO 19005 rules for PDF/A-1a through PDF/A-4f and reports pass or fail with the exact rule that failed. Never trust a converter's own claim without validating.
Yes — a PDF/A file is still a PDF, so any PDF editor can open it and save it as a normal PDF. You lose nothing by archiving in PDF/A, because the format is a subset of PDF, not a different file type.
PDF/A parts and conformance levels: ISO 19005-1 (2005), ISO 19005-2 (2011), ISO 19005-3 (2012), ISO 19005-4 (2020). Validator: veraPDF Consortium (veraPDF). PDF/A-2b as the most widely adopted standard: industry guidance, 2026. Conversion methods verified against LibreOffice, Microsoft Word and Ghostscript documentation. Researched by PDFBuddyFree Team · Last updated November 15, 2026.
Card photos: Jeremy Corbyn / Wikimedia Commons, CC BY 2.0 · Joe Mabel / Wikimedia Commons, CC BY-SA 3.0
Educational content only. Nothing on this site is legal or professional advice.