Epub Checker & Fixer

Find broken markup, packaging errors, dead links and invalid nesting in an Epub, then fix the mechanical problems with one click. Everything runs in your browser. Your book never gets uploaded.

Drop your .epub here, or click to choose.

Checked on your device. Nothing is uploaded.

The real rulesBuilt from the same Relax NG schemas the official Epubcheck validates with.

InstantNo Java, no install. A full novel checks in well under a second.

Repairs you can trustFixes are mechanical only, and the result is re-checked before you download it.

Your book never leaves your device. The check runs entirely in your browser using WebAssembly. Nothing is uploaded.

What this checks

This checker validates the things that actually stop Epubs opening or rendering correctly:

XML well-formedness: unclosed, mismatched or stray tags in every content file.
OCF packaging: the mimetype entry, container.xml, and a parseable OPF.
Manifest & spine: every manifest file exists and every spine reference resolves.
Links & resources: internal links and image/stylesheet references point to files that exist.
Content model: element nesting and attributes against the Epub and HTML5 rules (a <div> inside a <p>, for example), with XML namespace resolution and separate Epub 2 and Epub 3 rule sets.
Identifiers: id values must be valid XML Names, and the NCX identifier must match the package identifier.

How close is this to the official Epubcheck?

This is a real online Epub validator, not a lint pass. The content-model rules are generated from the same RELAX NG schemas the official Epubcheck uses, and the checker is developed by comparing its output against Epubcheck on real-world books. It is also fast. No Java, no install, no upload: a 25 MB novel validates in under a second. For certified conformance, including every Schematron rule and the accessibility checks, run the official Epubcheck as the final gate.

What "Fix safe problems" repairs

The fixer only makes mechanical repairs. It never rewrites your text:

Rebuilds a missing, misplaced or compressed mimetype entry.
Synchronizes the NCX dtb:uid with the package identifier.
Renames invalid id values (a Calibre export classic) and updates every internal link that points at them, so nothing breaks.

These three repairs are often all it takes to get a clean report; they cover the most common reasons a store or reader rejects a book. Problems inside the prose itself, like a <div> inside a <p> or stray text between paragraphs, are reported with the source line so you can fix them in your editor. Rewriting your content is a judgement call, so the fixer leaves it alone.

Common errors, explained

Guides to the validation errors people hit most, with fixes: PKG-006 (mimetype packaging), RSC-005 (content rule violations), RSC-007 (missing resources), RSC-016 (unclosed tags), and the rest. If a store bounced your book, start with KDP rejections or Apple Books rejections, and if you just converted it, see errors in Kindle-converted Epubs.