How to convert a PDF to ePub with reflowable text
By ScoutMyTool Editorial Team · Last updated: 2026-05-28
Introduction
A friend sent me their self-published book as a PDF and asked why the Kindle reader showed it as a tiny postage-stamp page no one could read. PDF is fixed pages; ePub reflows. The conversion sounds straightforward and is usually not, because converters take the path of least resistance and produce fixed-layout ePubs that defeat the whole purpose. Reflowable conversion requires a structured text source, careful column handling, real heading semantics, a hand-authored cover, and a validation pass. Here is the workflow that takes a PDF book and produces an ePub that reads cleanly on a Kindle, an iPad, and an Android phone in night mode.
Vocabulary, quickly
| Term | Meaning |
|---|---|
| Reflowable text | Text that re-wraps to fit screen size — the point of ePub |
| Fixed-layout ePub | ePub that pins pages to dimensions; not reflowable |
| EPUB 3 | Current spec; supports HTML5, CSS3, and accessibility metadata |
| OPF package | Container manifest listing every file inside the ePub |
| NCX / nav.xhtml | Table of contents formats — readers use one or both |
| Cover image | Required ePub asset; bookshelf views default to a placeholder otherwise |
| Reflow gotcha | Two-column PDF that flows columns end-on-end without column-detection |
Step by step
- Start from a tagged, single-column source if possible. Re-export the book PDF from the manuscript with proper heading styles. Saves hours of post-conversion cleanup.
- OCR scanned pages first. Reflowable conversion needs text; an image PDF can only convert to fixed-layout ePub.
- Convert with column-aware tool. Calibre with "treat as multi-column" hint, or a custom script that segments by column before extract.
- Audit the heading tree. Open the ePub, walk the HTML, confirm chapters are H1 and sections are H2. Rebuild manually where the converter missed.
- Wrap images and captions in FIGURE. One pass through the HTML files; takes 10 minutes for most books.
- Author a real cover. 1600x2400 JPG with title and author visible at thumbnail size. Substitute in the OPF manifest.
- Fill required metadata. Title, creator, language, identifier, modified-timestamp — all required by EPUB 3.
- Validate with EPUBCheck. Fix every error; warnings are negotiable but errors usually mean the store will reject the file.
- Side-load to a real reader. Kindle Previewer, Apple Books, Google Play Books Reader. Read the first chapter — if anything looks wrong on a 6-inch screen, fix it before publishing.
ePub conversion checklist
- Text reflows on a 6-inch screen at the reader\'s default font size — no horizontal scrolling.
- ToC opens to chapter starts, not to mid-chapter pages.
- Images scale within the page width; no horizontal overflow on small screens.
- Cover image renders at thumbnail size legibly; title and author readable in 200x300.
- EPUBCheck reports zero errors and only acceptable warnings.
- Night mode renders correctly — no light-on-light text where a CSS rule forced a color.
Related reading and tools
FAQ
- Why is my converted ePub one giant image per page?
- The converter treated the PDF as an image PDF and produced a fixed-layout ePub. Re-export from the source PDF with text-extract enabled, or run OCR first if the PDF is genuinely scanned. The converter needs a text stream to produce reflowable output; without it, the only honest conversion is fixed-layout, which defeats the purpose of ePub.
- My two-column PDF turned into nonsense — columns interleaved. Fix?
- You ran a converter without column detection. Use a converter that knows about multi-column layouts (Calibre with the "treat as multi-column" hint, or a script that segments the page by column before extracting text). One-column PDFs convert cleanly with any tool; multi-column PDFs need the hint or the output reads bottom-of-col-1 immediately followed by top-of-col-2.
- How do I keep chapter structure in the ePub?
- Tag PDF headings with proper heading levels before conversion — H1 for chapter, H2 for section. The converter reads the PDF tag tree and produces matching HTML headings, which the ePub uses to build the navigation. If the source PDF has no tag tree, you can post-process the ePub by find-replacing visual heading styles into H1/H2, but it is slower and error-prone.
- What about images and figure captions?
- Images convert as inline IMG tags; captions usually do not stay attached unless the PDF tag tree marked them as figure captions. Post-conversion, walk the ePub HTML files and wrap each image plus its caption in a FIGURE element. The reader announces the caption with the image and the layout stays sensible on a phone screen.
- Will the ePub pass EPUBCheck?
- Only if you build the metadata block carefully — title, creator, language, identifier, and modified-timestamp are required. Calibre and most converters fill these from the PDF Info dictionary; verify and edit if needed. Run EPUBCheck before publishing; the Kindle and Apple Books stores reject files that fail.
- How do I add a cover image?
- Designate a JPG or PNG as the cover in the OPF manifest with properties="cover-image". Most converters auto-generate a cover from the first PDF page; that is usually ugly. Hand-author a cover at 1600x2400 and substitute it in before validation. The bookshelf view in the reader is the first impression.
Citations
- Wikipedia — “EPUB — format, structure, validators.” en.wikipedia.org/wiki/EPUB
- Wikipedia — “Reflowable document — layout principles.” en.wikipedia.org/wiki/Reflowable_document
- Wikipedia — “Calibre (software) — conversion pipeline.” en.wikipedia.org/wiki/Calibre_(software)
Convert PDF to ePub in your browser
Extract text, rebuild structure, embed a real cover, and validate — ScoutMyTool runs the conversion locally so your manuscript never leaves your machine.
Open the PDF toolkit →