EPUB 2 or EPUB 3? That Is Not a Question

question mark on a teal background; decorative
  • Sumo

The word on the street is that the majority of ebooks submitted to various retailers are still native EPUB 2. Ben Dugas of Kobo reports that EPUB 2 makes up about 70% of incoming content.

Wait, what?

Let’s dissect this a little bit. It could be because developers are accustomed to their tools and are reluctant to change. As someone who learned how to make ebooks when there weren’t a lot of resources about how, I can empathize. It is a steep learning curve for someone who doesn’t come at this from a technical field like, say, web development. The old tools stills work, and the ebooks they produce aren’t rejected so why change?

It could be because developers are accustomed to their tools and are reluctant to change.

It could be that large companies can’t change very quickly. Or it might be that a well-established workflow is just too tricky to change when the benefits feel ethereal and unclear.

It could be that mostly self-publishers using cheap-to-free and out-of-date converters make up the bulk of these EPUB 2 producers.

I produce a trio of ebooks for Publisher’s Weekly several times a year which get distributed via Ingram to Netgalley, major retailers, Overdrive, and Edelweiss. For technical reasons we were required to stick to EPUB 2 until early 2016. But we have, thankfully, moved on to EPUB 3. I am happy to report that I haven’t personally touched EPUB 2 for eighteen months.

It could be that I am missing something big. What are the fundamental reasons to stick with EPUB 2? I would love to hear from these developers.

Some technical reasons to move on

Let’s crowd source a great big list of reasons to ditch EPUB 2 and upgrade your ebooks to EPUB 3. Please ping me on Twitter (@LauraB7) or suggest additions in the comments. I will update this article with your suggestions. I’ll start.

Let’s crowd source a great big list of reasons to ditch EPUB 2 and upgrade your ebooks to EPUB 3.

Rich Navigation

One of the main reasons to embrace EPUB 3 is it’s navigational capabilities. There is the traditional EPUB navigation – derived from the NCX in the bad old days – from the toc.xhtml. But you can add additional ways to navigate the ebook which can add layers of richness, especially for complex content. Lists of tables, figures, illustrations, maps – the possibilities are endless. You can also develop a print-corollary page-list so that print and digital consumers of an ebooks can literally be on the same page.

With this kind of rich navigation, your readers have a multitude of ways to get into your content.


In EPUB 3, you can embed schema.org metadata into the ebook’s OPF. By describing the accessibility modes, hazards (or lack thereof), and features, you can make your content more discoverable to people with specific reading needs.

A/V Content

Including audio and video content hasn’t really caught on widely in ebooks, partly because of the bulk they add to ebooks, partly because of a/v content no working broadly on the device spectrum, but if you want to include the odd video snippet or chunk of audio, you are much better served by EPUB 3.

ARIA Descriptions

For accessibility purposes, upgrading image descriptions to aria-describedby allows producers to create richer image descriptions to attach descriptive information to one or more elements through the use of an id reference. This is a web standard ported to ebooks bringing EPUB more inline with the accessibility best practices.

Creating MOBI files

A screenshot from the Kindle Publishing Guidelines pointing to the EPUB 3.0 specification.

Creating well-structured EPUB 3 files means that the process of converting to MOBI/KF8 is easier.  The EPUB 3 standard is referred to no fewer than five times in Amazon’s Kindle Publishing Guidelines. Those guidelines as much as say that following the navigation requirements of the EPUB 3 specification will lead to a better reading experience in Kindle products. You don’t want to hobble your MOBI files, do you?

16 Responses to “EPUB 2 or EPUB 3? That Is Not a Question”

  1. Rachel Comerford says:

    I definitely don’t want to hobble my MOBI files BUT I have noticed that when using coresource to push files to Amazon, Amazon is validating against ePUB 2.0 specs, not 3.0 – leading to errors and warnings being raised for files that are styled to ePUB 3.0 or 3.1. We still his the 3.1 standard but all that back and forth on files we already validated is hobbling my process timelines 🙁

  2. twobyte says:

    There are no reasons to “ditch” old and tested ePub 2. ePub Spec has always been providing forward as well as backward compatibility For example NCX can coexist with Navigational Document etc. Moreover I can suggest an app (Mac only) which will make ePub 2 file compatible with ePub3 and vice versa for that reason App is called Chapters and available on the Mac App store. Hope that helps with your discussion!

  3. Pablo Nolla says:

    Our workflow is ePub3 first,
    but then some retailers only validate ePub2… so we end up with two files, two ePub formats… + mobi of course…

    It would be great to have a single ePub3 file that would validate also on those reluctant retailers. Is that what Chapters do?

    using a single file

  4. twobyte says:

    Hi Pablo,

    Chapters is an editor for ePub files on Mac only for now. It can open ePub2 or ePub3 files no problem. It doesn’t validate against anything but instead it does its best to show up the file content. Once you begin editing ePub file (it doesn’t matter ePub2 or ePub3) even so slightly or you save as to a different file Chapters will use XSLT transforms on NCX in case ePub2 to produce a Navigation Document and adds it to manifest then it adds any required by ePub3 metadata adding it too to the package OPF and upgrades package version to 3. In case it was ePub3 file it does the same in the opposite direction – transforming Navigation Document into NCX adding it to manifest and adding metadata like coverImage for backward compatibility with ePub2. All of that is done transparently and by default. Chapters is not yet fully features as an editor though, but it is quite stable and reliable app.

  5. twobyte says:

    By the way, XSLT stylesheet for transforming ePub3 Navigation Document into NCX is written by Matt Garrish who has been on ePub committee for quite a few releases now. You can check out the stylesheet and use it from his blog at http://matt.garrish.ca/odds-n-ends/ncx-generator/ Another XSLT stylesheet for the opposite transformation from NCX to Navigation Document was written by Keith Fahlgren from O’Reilly. Chapters acknowledges using both stylesheets as they were open sourced by the authors. And last but not least, I’d like to urge anyone working with ePub for living to read spec itself which is very well written and quite approachable. Hope that helps!

  6. Nick R says:

    My experience is that the vast majority of ebooks being sold are novels… So, all those features don’t really make a hill of beans difference. I read a ton of ebooks. The only “features” I’ve ever used are next page and previous page… Not 100% true, I use Kindles feature to define and translate every now and then.

    Take correlating print and ebook pages. This strikes me as a friend case, who’s out there reading both copies? I’d love to see some real use case studies showing how real ebook readers use ebooks.

    I’ve always suspected that most of these features came about from print advocates trying to claim print was better than the ebook. It’s not. Which is clear once you’ve tried both. Why do retailers require a toc in a novel? What physical novels ever have a toc?

  7. Rachel says:

    As someone who publishes in the educational space, I can tell you that these features have made an incredible difference in the quality and navigability of our titles. But more than that – these features make a huge difference for users that are reading using assistive technology. The accessibility metadata provides users with an overview of what they can expect to work and not work for them. It also reveals hazards – like flashing – in order to protect users.

    In terms of a TOC – I have yet to meet a child that isn’t more adept with a screen than I am. Among those novels sold are children’s “chapter” books, including those assigned for school reading. The TOC is a necessity for these learners.

    Finally, in general, these updates mean a greater flexibility in the ebooks that are produced – we can use these standards to create ebooks that are far more fun to read than the ones where you’re just hitting next and previous.

  8. Felipe says:

    I like so much of Sigil. This free ePUB editor is very powerful and with plugins is possible convert ePUB 2 to ePUB 3
    Very nice

  9. Nick R says:


    I’m not arguing with the benefits of the standard. I like the improvements. I can totally see the need for them in nonfiction. I was trying to explain why I thought adoption was so poor. I find children’s books to be poorly representative in the ebook market. And to my point physical chapter books don’t have a TOC nor should they.

  10. Rachel says:

    Thanks for clarifying Nick! I read your comment as a criticism of the necessity of the changes so it was good to hear your POV. I will continue to challenge you on the TOC question though – most children’s chapter books do have TOCs and should – it’s an educational tool.

    Anyway, the idea that constructing a TOC (even one as simple as a novel’s – Chapter 1, Chapter 2, Chapter 3) would prevent publishers from moving to the new standard seems a little extreme to me. Especially as someone who manages standards for a publisher!

    There’s no question that these can be a challenge to adhere to but what holds publishers back, in my experience is one of two things:
    1) Lack of awareness – not getting involved in the standards groups can mean roll out of new file types can be very slow
    2) Lack of reader compatibility – basically, we have to keep building multiple versions of files to accomodate readers like the Kindle that are validating against an old standard while they update their own internal standard.

    So why did we make the change? In a broad sense – this provides a better reader experience, especially in the educational field. Faster and clearer navigation; easier to assign readings to students that may be using print, digital, or some combination of the two; clarity into accessibility of content; and finally, the flexibility to create a more interactive reading experience.

  11. When I read an ebook, it’s the same as when I read a print book: I want to read, not watch a video or even look at pictures. My mind can create vivid images from the words. For me, that’s the joy of reading.

  12. Scott says:

    RE: “Take correlating print and ebook pages. This strikes me as a friend case, who’s out there reading both copies?”

    1. In a college literature class some students may have the ebook version, others the print edition. It’s pretty cool if a professor can discuss a passage on, say, page 49, and all students can go to that page.

    2. Corresponding print and ebook page numbers is also a potentially great advantage for academics and researchers who could more easily cite an edition of a work regardless of whether they found the info in the print or ebook version.

  13. Robert Nagle says:

    I can’t believe no one has mentioned the obvious — css media queries! (Of course, it was probably a bigger deal when Kindle announced their support for media queries). But it totally changed the way I made ebooks.

  14. I’ve been struggling for two months to convert my nonfiction, self-published (274 pp) book and workbook (110 pp) to an ePUB 3. It’s been a nightmare. A professor at a nearby community college has even given it a go. Neither of us have been able to get the manuscript to look right. Why on earth does it have to be so capricious and complicated? Both were created within a single document. Both are published and available in print on Amazon. But STILL I have no eversions. I’ve watched YouTube videos until my eyes bled. Still, no eversions. The Table of Contents is a mess, the page breaks are ineffective. The author has pulled all her hair out and is wimpering pathetically. Please- what am I missing here? Any videos or instructional books you would recommend?

  15. Steve says:

    Funny that Kobo positions itself on the question when none of their devices actually correctly handle ePub 3 correctly. They don’t even want to hear about it.
    ePub3 files shown on Kobo devices skip all blank lines, especially between paragraphs as the rendering engine doesn’t correctly handle this. Tech support just told me I had to reboot the device. The most stupid answer I ever got in my almost 40 years in computing. Just like if a reboot was related to the algorithm involved in rendering ePub 3 files on screen.
    Similar ePub 3 files display correctly, as they were designed on any other devices, in any other software like Books on macOS, iOS, on ePub readers on Windows and Android, including Calibre.
    They just don’t want to hear about the problem.
    The main difference between ePub 2 and 3 is a better use of CSS to format the text. Something they obviously didn’t take into consideration at Kobo.

  16. Kate says:

    You very likely have found a solution by now, but here’s what I did with a collection of illustrated short stories with complicated formatting. I made a simplified ebook doc in Word created from the original doc (Word) to Draft@Digital, downloaded it, and through trial and error was able to make a good-looking epub file.

    I also did the same with Kindle Create, but that results in a file that can only be used at Amazon. Kindle Create allows you to design both fixed format books for complicated books like I assume you have, or to tinker with them and turn them into a simpler, flowable style. Both options require A LOT of trial and error.

    I wasn’t able to pay an ebook conversion service and am pleased that I have now have three decent ebooks out of this epic struggle.