Veröffentlicht
A Plain Guide to EPUB 3 — What It Adds, What It Breaks
An EPUB 3 file looks the same as an EPUB 2 file from the outside. Underneath, the format changed more than most readers realize — new capabilities, and a few old ones quietly removed.
What EPUB 3 Is
An EPUB file is a ZIP archive with a few required parts: a package document that lists every resource and defines the reading order, a navigation document that works as the table of contents, and the actual content — pages of XHTML, plus whatever images, fonts, or audio they reference. EPUB 3 built all of this on ordinary web technology — HTML5, CSS, SVG — rather than a set of formats invented just for ebooks.
That choice matters more than it sounds. A plain-text or plain-markup format tends to outlast anything proprietary, because the tools that read it are the same tools that read the rest of the web. EPUB 3 leaned into that instead of away from it.
What It Adds
The headline change is the base markup: EPUB 3 moved from XHTML 1.1 to HTML5, bringing semantic elements like section, nav, and aside. That sounds technical. In practice it means a table of contents and chapter structure that reading systems can parse more reliably.
Three additions matter most for actual reading:
- Fixed Layout Documents. EPUB 2 really only handled reflowable text — paragraphs that resize to fit the screen. That doesn't work for comics, picture books, or anything where the page layout is part of the content. EPUB 3 lets a publisher pin exact page dimensions and spread position, closer to how a PDF behaves, but still inside an EPUB container.
- MathML. Equations can now be structured data a screen reader can interpret, instead of a flattened image with no readable text underneath — a real gain for technical and academic material.
- Media Overlays. Pre-recorded narration can sync to the text itself, which helps both accessibility and read-along use.
EPUB 3 also widened font support (OpenType and WOFF, with obfuscation options for licensed fonts) and added proper support for bidirectional text, ruby annotations, and mixed character sets — useful for anything that isn't left-to-right Latin script.
What It Breaks
Additions are the easy part to like. The harder part is what EPUB 3 removed outright to get there.
The NCX — EPUB 2's proprietary table-of-contents format — is gone, replaced by the Navigation Document. The guide element was superseded by navigation landmarks. And a handful of EPUB 2 mechanisms disappeared entirely: DTBook as an alternate content syntax, the "Filesystem Container" option (EPUB 3 files must be ZIP archives, full stop), the tours element, and support for XML 1.1.
Validation got stricter too. Manifest fallbacks are now restricted to spine documents only, filenames inside the archive follow tighter character rules, and EPUB 3 requires Unicode 5.0.0 where EPUB 2 only required Unicode 2.0. None of this is dramatic by itself. But it means an older EPUB 2 file built around any of the removed features can fail EPUB 3 validation, or render incorrectly in newer software, even though it opened fine for a decade.
Where You'll See It
Adoption is close to universal at this point. Apple Books and Kobo support both EPUB 2 and EPUB 3, Google Play Books lets readers download purchased titles as EPUB, and in 2022 Amazon's Send to Kindle service dropped support for its own proprietary Kindle format in favor of accepting EPUB directly — a notable concession from a company that spent over a decade pushing its own format instead.
Where you'll actually run into EPUB 3 day to day: sending yourself something you downloaded from the Internet Archive, or opening a scanned document that needed OCR before it behaved like a proper EPUB rather than a stack of page images.
How Margina Handles It
Margina opens EPUB 2 and EPUB 3 files the same way — no conversion step, no account, no upload. You send yourself the file and read it. There's no server processing it on the way in, and nothing about what you're reading gets tracked. If you want a copy on the App Store or Google Play, that's where to find it.
Frequently Asked Questions
Is EPUB 3 backward compatible with EPUB 2?
Mostly for reading, not always for authoring. Modern reading systems render EPUB 2 files fine, and an EPUB 3 file that avoids the removed features will open in most EPUB 2-era software too. But EPUB 3 formally removed several EPUB 2 mechanisms — DTBook as an alternate syntax, the Filesystem Container option, the tours element, and XML 1.1 support — so an older file built around any of those can fail EPUB 3 validation or misrender in newer tools.
Do I need to convert my old EPUB 2 files?
Usually not. Reading apps built for EPUB 3, including Margina, still open EPUB 2 files directly. Conversion only matters if you're editing or re-publishing a file and want to add EPUB 3-only features like fixed layout, MathML, or media overlays — none of which exist in the EPUB 2 spec.
Why did EPUB 3 replace the NCX table of contents?
The NCX was a proprietary EPUB-only format. EPUB 3 replaced it with the Navigation Document, a plain XHTML file using the standard HTML nav element — so the table of contents is built from the same web technology as the rest of the book, and reading systems can parse it with an ordinary HTML parser instead of a special-purpose one.
Does EPUB 3 support comics and fixed-layout books?
Yes — one of its biggest additions. EPUB 2's layout model was reflowable text only, which doesn't work for comics, picture books, or anything where exact page geometry matters. EPUB 3's Fixed Layout Documents let a publisher pin exact page dimensions, orientation, and spread position, similar to a PDF but still inside the EPUB container.
Conclusion
EPUB 3 is EPUB 2 rebuilt on modern web standards, plus a short list of things it was willing to leave behind to get there. Most of that is invisible day to day — a file you already own will keep opening the same way it always has. It only becomes visible the moment an old file leans on something the format quietly retired.