10 EPUB Formatting Mistakes That Get Books Rejected

Many authors add blank lines in their manuscript editor expecting them to become spacing in the final ebook. EPUB readers handle empty <p></p> tags inconsistently — what looks right in Calibre can render as random blank gaps on Kindle and Kobo devices. Paragraph spacing belongs in CSS margin-bottom on the paragraph style, not in blank lines in the manuscript text.

Most EPUB converters build each chapter from the same structural template and copy heading and section IDs into every chapter file. EPUB requires every id attribute to be unique across the entire publication — duplicate IDs cause EPUBCheck errors and Apple Books rejections. The error is almost never visible in reading-app previews, which is why it makes it past authors who do preview their files.

The NCX file (toc.ncx) provides chapter-level navigation to reading systems and is required even when your converter claims to produce EPUB 3. When it's missing or references the wrong chapter anchors, KDP and Apple Books lose the ability to provide in-reader chapter navigation — and some distribution pipelines reject the file entirely on submission rather than listing it with broken navigation.

EPUBCheck passes files where the cover image is listed in the manifest without properties="cover-image". Platforms don't. Amazon KDP and Apple Books use this property to identify which image to display as the storefront thumbnail. A missing property means your cover either doesn't appear in the store or appears as a blank gray box — even though the file itself uploaded without error.

The OPF spine controls the order chapter files are presented to reading systems. If the spine order diverges from the intended reading order — usually because chapters were reordered after export without updating the spine — the resulting ebook reads the wrong chapters in the wrong sequence. The cover and TOC look correct; the problem only appears when a reader navigates chapter-by-chapter.

Changing a chapter filename after the OPF manifest is generated leaves a manifest item pointing to a file that no longer exists. EPUBCheck flags this as RSC-007. It happens most often when chapters are reorganized after a first export, or when a converter generates chapter files with different names than it registers in the manifest. It's invisible on a quick look at the final EPUB but blocks Apple Books submission.

Tools like older Calibre versions and some Word converters output EPUB 3 files with EPUB 2 metadata attributes (opf:role, opf:file-as) on metadata elements that no longer support them. The file declares EPUB 3, EPUBCheck validates against EPUB 3 rules, and the mismatch throws RSC-005 errors — errors that weren't present in the original EPUB 2 source the author shipped fine from in a previous edition.

EPUB chapter files are XHTML, not HTML — the & character must be written as & in all contexts. A literal & in a title, heading, or book description embedded in XHTML causes a parse failure at the XML level. This is one of the most common errors in files converted from DOCX, where ampersands appear naturally in text and converters don't always escape them correctly.

XHTML requires all elements to be explicitly closed. Void elements like <img> and <br> need a self-closing slash (<img />, <br />). HTML5 allows these to be left open, so files hand-edited with web development habits — or pasted from blog HTML — frequently fail XHTML validation in ways that EPUBCheck catches but browser-based preview tools don't surface.

Every EPUB requires a dc:identifier element, and the OPF <package> element's unique-identifier attribute must point to that element's id by name. When they don't match — usually because a template was edited or an identifier element was moved — EPUBCheck throws OPF-048. Apple Books rejects the file outright; KDP accepts it but may show metadata warnings in the dashboard.