PDF/A is a family of ISO standards for a restricted form of PDF, designed so that a document keeps looking exactly the same for the long term.
Plain PDF is flexible. It can pull fonts from the computer that opens it, play video, run scripts and lock itself with a password. That flexibility is great today and a risk in twenty years, when the software that understood a particular trick may be gone. PDF/A removes the risky parts on purpose.
The "A" stands for archival. You'll run into the term when a court, a records office, a school or a government form asks for "PDF/A" files, and almost nowhere else. For ordinary paperwork at home, a regular PDF does the job. Still, it's worth knowing what the label promises and what it doesn't.
What a PDF/A is
The PDF Association, the industry group that supports the standard, describes ISO 19005 as defining a PDF-based format that preserves the static visual appearance of electronic documents over time, independent of the tools and systems used to create, store or display them. It does that by listing which PDF features may be used, and restricting how.
The Library of Congress, which tracks formats for its own long-term collections, says PDF/A tries to maximize three things: device independence, self-containment and self-documentation. In practice, that means the file carries everything it needs to be displayed, and it describes itself with standard metadata.
The PDF Association's FAQ adds an honest caution: PDF/A is "just another file-format" and gives no special protection against data corruption. It makes a file easier to read in the future. It doesn't make a file immune to a damaged drive, so you still need backups.
What's on it
PDF/A comes in parts, each based on a different PDF version, and each with conformance levels. The Library of Congress describes the constraints common to the first three parts.
| Part or rule | What it means | Source |
|---|---|---|
| PDF/A-1 (ISO 19005-1, published 2005) | Based on PDF 1.4; levels A and B only; no transparency | ISO; PDF Association |
| PDF/A-2 (ISO 19005-2:2011) | Based on PDF 1.7; can embed PDF/A-compliant attachments | Library of Congress |
| PDF/A-3 | Adds permission to embed files in any format | Library of Congress |
| Fonts | All fonts must be embedded, and legally embeddable | Library of Congress |
| Forbidden content | Audio, video, 3D artwork, JavaScript and executable file launches | Library of Congress |
| Color | Color spaces must be specified in a device-independent way | Library of Congress |
| Encryption | Disallowed | Library of Congress |
| Metadata | Standards-based metadata is mandated | Library of Congress |
Conformance levels, for the first three parts, are A, B and U. Level A meets every requirement of the standard. Levels B and U still guarantee the visual appearance but ask less about structure and meaning. The Library of Congress notes that level B is the level typically used for PDF/A files created from scanned pages. The PDF Association's plain-English names for the levels are Accessible, Basic and Unicode.
What to keep and for how long
PDF/A is a format, not a record, so retention follows the document, not the file type. How long you keep a lease, a tax return or a warranty depends on the document and on your situation. Our guide to what documents to keep and for how long is the place to start, and the glossary entry on document retention explains the idea.
A few practical points follow from the table:
- Keep a master. If you convert a PDF to PDF/A, keep the original too, in case the conversion changes anything you care about.
- Don't expect a password. Encryption is disallowed, so if a document is sensitive, protect where it's stored. See encryption at rest.
- Scans are pictures. A PDF/A built from scanned pages preserves the image. Searchable text is a separate matter.
- Ask what's actually required. If an agency says "PDF/A," check which part and level it wants. If you can't tell, ask them.