The name is short for Electronic PUBlication. Knowing what is inside gets you further than the name, though, so here is the conclusion first.
The inside of an EPUB is almost the same thing as a web page.
You may have heard that web pages are made of HTML and CSS. EPUB is exactly the same: the text is HTML, the appearance is CSS. Add images and fonts, pack the lot into a single file, and that is an EPUB.
So EPUB is not some entirely new, special technology. It is the technology of the web, reassembled into the shape of a book. That is why it reads on devices from any manufacturer, and why anyone can build one from the published specification.
If all you want is to hand out an ebook, PDF looks like it would do. Plenty of books are distributed as PDF to this day.
So why did EPUB appear? Because people read in wildly different environments.
PDF freezes "how it looked when printed on paper". A document made at A4 is still A4 when you open it on a phone. Fit it to the screen and the type shrinks; zoom in and you have to scroll sideways.
EPUB differs here. The text is poured in to fit the reader’s screen size and type size. You can read on a phone in large type as comfortably as reading a paperback. That capacity to reshape itself around the reader is exactly why EPUB became the standard for ebooks.
In a paper book, "line 3 of page 40" is the same place for everyone. In EPUB, where pages break varies from reader to reader, so "which page" is not a fixed idea. That is not a flaw — it is the flip side of designing around the reader.
It is not widely known, but an EPUB file is a ZIP file.
Try changing the extension of an EPUB from .epub to .zip and unzipping it. No error; the contents simply unpack. Inside, you find files like these.
Folder names vary by the tool that built it, but the idea is the same. Text, images and design instructions packed into one box together with a contents list and a manifest. That is all an EPUB is.
The important one is content.opf , which holds the book’s title and author, the list of files it contains, and the order pages are turned in. A reading app looks at this file first to understand the book’s structure.
What each of those files does is covered in a separate article.
There is one distinction to learn before anything else.
| Shape | Character | Books it suits |
|---|---|---|
| Reflowable | Text flows to fit the screen and the type size | Novels, essays, practical books |
| Fixed layout | The appearance of the page is locked | Manga, photo books, picture books, magazines |
The "reshapes itself around the reader" strength described above applies to reflowable. It suits books where the text is the star.
Manga and photo books, on the other hand, have an intention — "this page should look like this" — and reshaping ruins it. That is what fixed layout is for: you set the page size and place images or text on it.
Same name, but the two are built quite differently inside. Deciding which one your book is, first, is step one of making an EPUB.
Because EPUB is an open common specification, plenty of apps and devices support it: Rakuten Kobo, Apple Books, the apps of various ebook stores, free desktop readers, and more.
There is one large exception.
Amazon uses its own file format. You can submit an EPUB to KDP (Kindle Direct Publishing), but it is converted internally to Amazon’s own format on the way in. What readers read is the converted file.
This is the single biggest place people trip up. It displays correctly in your own reader, then looks different once it is on Kindle — that happens. With Japanese vertical text especially, the difference tends to surface.
So if Kindle is where you are publishing, a step that checks the real appearance in Amazon’s own tool (Kindle Previewer) is essential. However correct the EPUB is before conversion, that alone is not enough.
The EPUB specification is maintained by W3C (World Wide Web Consortium), the international body that sets web standards. It used to be drafted by IDPF, an electronic-publishing body, and has since passed to W3C.
The version numbering is where it gets slightly confusing.
What is in wide use is "EPUB 3", but the specifications carry finer numbers: EPUB 3.2, EPUB 3.3. EPUB 3.3 was published as a W3C Recommendation on 25 May 2023 and was the first edition in the EPUB series to become a formal W3C Recommendation. Compatibility with EPUB 3.2 is preserved: a file that is valid as EPUB 3.2 is defined to be valid as EPUB 3.3.
W3C’s own explanation notes that the difference between EPUB 3 as a "format" and the numbers on the specification (3.2, 3.3 …) is a common stumbling block for newcomers. The decimal is there to distinguish editions of the specification; a new number does not mean a different format.
As of August 2026, the next edition, EPUB 3.4, is in progress. A Candidate Recommendation was published on 21 July 2026, and a formal Recommendation is expected on or after 19 October 2026.
Items under consideration for 3.4 include dark-mode support and standardising footnotes and endnotes, and there is also a dedicated team looking at digital comics. The open questions about handling manga in EPUB are being argued out internationally right now.
Once you know the insides are HTML and CSS, you may think you could just write it yourself. Technically you can, but in practice there is a list of fine rules to obey.
mimetype first in the ZIP, uncompressedcontent.opf , without exceptionMiss any one of them and the book fails to open in a reading app, or gets bounced at store submission. Assembling all of this correctly by hand, every time, is hard work even when you are used to it.
So in practice people use a tool that takes a manuscript or images and assembles the EPUB for you. EPUB FACTORY is one of those. It runs entirely in the browser, and your manuscript is never sent to a server.
Pick a folder of images or a text manuscript and you get an EPUB 3 ebook file. No account, no installation.
Try EPUB FACTORY