An EPUB is a reflowable ebook file: a format designed to adapt to different screens, font sizes and reading devices rather than to hold fixed pages. EPUB 3 is the current standard, published by the W3C, and it is what the major ebook stores and reading apps use.
If you have written a book and want to sell it as an ebook, the EPUB is almost certainly the file you need to make. This page is about what it is. If you already know and want the internals, what's inside an EPUB file takes one apart.
What "reflowable" means, and why it matters
A printed page is fixed. The text was set once, at one size, on a page of one shape, and it stays that way.
An EPUB has no pages in that sense. It has text with structure — this is a chapter title, this is a paragraph, this is a scene break — and the reading device decides how to lay that out for the screen it has and the type size the reader chose. Turn the font up and the book gets longer. Rotate a tablet and every line re-wraps. Open the same file on a phone and on a 10-inch e-reader and you get two completely different sets of pages from one file.
This is the entire point of the format, and it is why the question "what page is that on" has no answer in an ebook. It is also why a PDF makes a poor ebook: a PDF is fixed pages, so on a phone it is either unreadably small or scrolled sideways.
Where EPUB is used
Nearly everywhere ebooks are sold or lent:
- Apple Books, Kobo, Google Play Books — EPUB is the upload format, and the only one.
- Amazon KDP — accepts EPUB alongside Word and a few others, and converts it to Kindle's own delivery format for readers. What you upload is EPUB; what the Kindle stores is Amazon's business.
- Libraries — OverDrive, Libby and most library platforms lend EPUBs.
- Aggregators like Draft2Digital, which distribute one EPUB to many stores.
The practical consequence: make one good EPUB and it goes everywhere. That is the argument for building the EPUB even if Amazon is your first stop.
EPUB 2 and EPUB 3
Two versions are in circulation.
EPUB 3 is the current standard — a W3C Recommendation, most recently revised as EPUB 3.3. It is built on modern HTML and CSS, supports proper navigation, accessibility metadata, embedded fonts, audio and video, and multiple writing directions. This is what you should produce.
EPUB 2 is the 2007 version. It still opens on everything, and some older devices and a few library platforms prefer it. It lacks most of what EPUB 3 added, but for a book that is entirely words the difference is invisible to the reader.
Most producers ship EPUB 3 with the EPUB 2 navigation file included as a fallback, so old devices get a chapter menu too. There is no reason to build an EPUB 2 on purpose unless a specific platform asks for it.
What is actually in the file
An EPUB is a zip archive with a strict internal layout. Rename book.epub to
book.zip and you can unzip it. Inside you find:
- one XHTML file per chapter — your actual text, as strict HTML,
- a stylesheet, which the reader is free to override,
- a package document listing every file, the reading order, and the book's metadata — its title, author, language and identifier,
- a navigation document: the list of links that becomes the chapter menu.
That last one is the thing most worth knowing about. The chapter menu on a reader's device is not generated from what the chapters look like. It is a literal list inside the file, and if a chapter is not in that list, it is not in the menu. This is the single most common way a home-made EPUB disappoints: it opens fine, reads fine, and has no navigation.
The full anatomy, with a real file unpacked, is in what's inside an EPUB file.
How to open one
Any reading app: Apple Books on a Mac or iPhone, Google Play Books, Kobo's app, Calibre on any desktop, or Thorium Reader if you want a free desktop app that follows the standard closely. Modern Kindles will also open an EPUB sent to them by email or through Send to Kindle, converting it on the way.
To look at one as a producer rather than a reader, unzip it as above and open the XHTML files in a browser.
Valid, and merely openable
An EPUB that opens is not the same as an EPUB that is valid. Stores run a validator on upload — EPUBCheck, the reference implementation maintained by the W3C and the DAISY Consortium — and reject files that fail it, sometimes with an error message and sometimes without.
Common failures are dull: a file listed in the package that is not in the archive, an unclosed tag in one chapter, a missing language declaration. None of them stop the book opening in an app, which is why authors are surprised when a store rejects a file that "works". Run EPUBCheck yourself before uploading anywhere. What EPUBCheck errors actually mean explains the messages.
How to make one
The realistic routes, for a book that started life in Word:
Export from Google Docs or a word processor. Google Docs will download a document as EPUB. The result opens, but you get no control over the navigation or the metadata and no validation, and the chapter menu is only as good as your heading styles.
Build it with a converter. Calibre and Pandoc will both turn a .docx into an EPUB. Both do exactly what the document's structure tells them — a chapter title styled as a heading becomes a chapter; one styled as big bold text does not.
Use a formatter that reads the structure. This is what this site does: it works out which lines are chapter openings and scene breaks, even where the Word file was styled by hand, builds the EPUB 3 with a real navigation document and metadata, and runs EPUBCheck on it before you download.
The tools, if you want them: create an EPUB from a Word manuscript, what a straight Word-to-EPUB conversion loses, validate an EPUB you already have, and one EPUB for Apple, Kobo and Google.