PDF to EPUB conversion produces broken ebooks because the two formats use fundamentally different layout models. PDF is a fixed-layout format that locks every element to an exact position on the page, while EPUB uses reflowable HTML and CSS that adapts to different screen sizes. Converting between them forces a translation that almost always fails, resulting in garbled text, misplaced images, and lost paragraph structure. The correct approach is to create your EPUB from the original source file, not from a PDF.
- What Is the Difference Between Fixed Layout and Reflowable Content?
- Why Does PDF Data Fail to Translate into EPUB Structure?
- What Goes Wrong During a PDF to EPUB Conversion?
- What Do KDP, IngramSpark, and Draft2Digital Actually Require?
- How Should You Create an EPUB from Your Manuscript?
- Can You Salvage a Broken PDF to EPUB Conversion?
- Frequently Asked Questions
What Is the Difference Between Fixed Layout and Reflowable Content?
PDF and EPUB handle page layout in opposite ways, and that difference is the root cause of every conversion problem.
PDF (Portable Document Format) is an ISO open standard (ISO 32000) designed to preserve the exact visual appearance of a document regardless of which device or software opens it. Every character, image, and line sits at a precise coordinate on a page of fixed dimensions. This is what makes PDF ideal for print: the file you export is the file your printer reproduces.
EPUB, the current version of which is EPUB 3.3 (a W3C Recommendation), is built on web technologies. Its content documents are XHTML, styled with CSS, and the entire package is a ZIP archive. Because the content is reflowable, it reflows to fit whatever screen the reader uses; a paragraph that fills half a page on a tablet might spread across three screens on a phone. The reader can also change the font, font size, and margins, and the text adjusts dynamically.
When you ask a conversion tool to turn a PDF into an EPUB, you are asking it to reverse-engineer a fixed, pixel-positioned document into semantic, reflowable markup. The two formats were never designed to be interchangeable.
Why Does PDF Data Fail to Translate into EPUB Structure?
The core issue is that PDFs do not store semantic information about your document’s structure. A PDF knows that a particular string of characters appears at coordinates (72, 640) on page 14 in 11-point Garamond. It does not know whether that string is a chapter heading, a body paragraph, a footnote, or a running header.
EPUB, by contrast, requires exactly this kind of structural information. An EPUB content document uses HTML tags to define headings (<h1>, <h2>), paragraphs (<p>), lists, block quotes, and other elements. These tags tell the reading system how to render the content and allow screen readers to navigate the document correctly.
A conversion tool looking at a PDF has to guess which text is a heading and which is body copy, usually based on font size and weight. It has to guess where paragraphs begin and end, often incorrectly merging lines that happened to sit close together or splitting paragraphs at page boundaries. This guesswork is inherently unreliable, and it gets worse the more complex your layout is.
What Goes Wrong During a PDF to EPUB Conversion?
The formatting failures from PDF to EPUB conversion tend to follow predictable patterns. Understanding them helps explain why cleanup is so difficult.
Broken line and paragraph structure
PDFs store text line by line. A conversion tool often treats each line as a separate paragraph, inserting hard line breaks where the original text simply wrapped to the next line. The result is an ebook where every line ends prematurely, making the text unreadable on wider screens.
Headers and footers mixed into body text
Running headers, footers, and page numbers in a PDF sit on every page as positioned text. Conversion tools frequently pull these into the main content flow, so your chapter text is interrupted every few paragraphs by the book title, author name, or a stray page number.
Displaced or missing images
Images in a PDF are anchored to specific page coordinates. EPUB’s reflowable model requires images to sit inline with the content flow. Conversion tools often place images in the wrong location, strip them entirely, or produce them at incorrect dimensions. For platforms like IngramSpark and Apple Books, each image in an EPUB must also stay under 5.6 million total pixels.
Lost formatting and styling
Bold text, italics, indentation, drop caps, and block quotes are all positional or font-based in a PDF. The conversion tool may not detect them at all, leaving you with a flat wall of unstyled text. If your PDF used embedded fonts, those fonts are almost certainly not carried across correctly either.
What Do KDP, IngramSpark, and Draft2Digital Actually Require?
Each major ebook distribution platform has its own file requirements, and none of them recommend PDF as a source for ebook content.
Amazon KDP accepts EPUB files that comply with the Kindle Publishing Guidelines. KDP also accepts Word documents (DOC/DOCX), which it converts internally. While KDP technically accepts PDF uploads for print interiors, using a PDF as the source for an ebook listing produces the same reflowable conversion problems described above.
IngramSpark requires EPUB 2 or EPUB 3 format for ebook distribution, with reflowable text preferred and a maximum file size of 100 MB. They do not accept PDFs for ebook distribution.
Draft2Digital is the most explicit: they do not accept PDFs at all for ebook conversion. Their accepted input formats are Word (DOC/DOCX), ODT, Rich Text, or a pre-formatted EPUB. This policy exists precisely because PDF to EPUB conversion produces unreliable results.

How Should You Create an EPUB from Your Manuscript?
The correct approach is to create your EPUB from the same source file you used to write the book, not from a PDF export of that file. Your Word document, Google Doc, Scrivener project, or InDesign file contains the semantic structure (headings, paragraphs, styles) that EPUB needs.
If you wrote your book in Word, you can upload the DOCX file directly to KDP, Draft2Digital, or use a formatting tool like Vellum, Atticus, or Reedsy to generate a clean EPUB. If you used InDesign for your print interior, export the EPUB directly from InDesign using its built-in EPUB export function rather than exporting a PDF and converting it afterward.
The key principle is simple: keep the PDF for print and create the EPUB separately from your source manuscript. These are two different output formats from the same source material, not one format converted into the other. Once you have your EPUB, validate it with EPUBCheck before uploading to any retailer to catch structural errors before they reach readers.
Can You Salvage a Broken PDF to EPUB Conversion?
If the original source file is genuinely lost and a PDF is all you have, limited cleanup is possible, though it is time-intensive and the results rarely match what a proper source-to-EPUB workflow produces.
Calibre, the open-source ebook management tool, includes a PDF to EPUB conversion engine with a Heuristic Processing option. When enabled, it attempts to detect and remove unwanted headers, footers, and page numbers using pattern matching. For example, it can strip repeated text that appears on every page and remove sequences that look like line numbers. These heuristics help, but they are imperfect; they work best on simple, text-heavy documents with minimal formatting.
For anything more complex (illustrated books, poetry, textbooks, cookbooks), a PDF-only starting point typically requires manual reconstruction. That means copying the text out of the PDF, pasting it into a word processor or ebook formatting tool, reapplying all heading styles and paragraph formatting by hand, and re-inserting images at the correct locations. At that point, you are effectively recreating the book from scratch rather than converting it.
The bottom line: if your source file still exists in any editable format, always use that instead of attempting to rescue a PDF conversion.
Frequently Asked Questions
Can Calibre produce a good EPUB from a PDF?
Calibre can produce a usable EPUB from simple, text-heavy PDFs, but the output almost always requires manual cleanup. Its Heuristic Processing feature helps strip headers and fix line breaks, though it struggles with complex layouts, images, and multi-column designs. For anything beyond a straightforward novel manuscript, the results are unlikely to meet retail quality standards.
Why does my PDF look fine but the converted EPUB is unreadable?
Your PDF looks fine because it is a fixed-layout format that preserves exact positioning. The EPUB looks broken because the conversion tool had to guess at your document’s semantic structure (headings, paragraphs, image placement) from positional data alone. That guesswork introduces errors that compound throughout the document.
Do any ebook platforms accept PDF uploads for ebooks?
Amazon KDP allows PDF uploads for print books, but for ebook (Kindle) listings, KDP converts your uploaded file into its own format. Uploading a PDF for an ebook listing typically produces the same reflowable conversion problems. IngramSpark and Draft2Digital do not accept PDFs for ebook distribution at all. All three platforms recommend EPUB or Word as the ebook source format.
Is a fixed-layout EPUB the same as a PDF?
No. A fixed-layout EPUB uses HTML and CSS to position elements on pages of defined dimensions, similar to a PDF’s visual result but built on web standards. Fixed-layout EPUBs are used for children’s picture books, comics, and heavily designed content. They are not the same as PDFs and require purpose-built creation tools; you cannot reliably produce one by converting a PDF.
What file format should I use to create both a print book and an ebook?
Start from a single editable source file such as a Word document, Scrivener project, or InDesign file. Export a print-ready PDF from that file for your paperback or hardcover. Separately, export or generate an EPUB from the same source for your ebook. This parallel-output approach avoids the PDF to EPUB conversion problem entirely.
