Free PDF to EPUB converter The right format,
ready for your reader.

Convert2EPUB uses credits for processing that costs more to run, rather than charging for each book. Ordinary text PDFs are free with Standard, and so are the first 10 recovered or OCR pages of every book. Past those, recovering an already-scanned text layer uses 1 credit per 10 pages and fresh OCR uses 1 credit per page. Pro uses 2 credits for every PDF page. You see the exact total before conversion starts, and failed conversions return reserved credits. If a book needs less work than quoted, unused reserved credits return to your balance. Translation costs 1 credit per 500 words. Scanned PDF pages are estimated from readable pages in the same file, or at 250 words each when the PDF has no readable text. Standard adds OCR credits when needed; Pro recognition is charged at 2 credits per PDF page.

New accounts receive 30 credits. Choose 250 to 2,500 credits on the slider, as a one-time credit pack or as a monthly plan. Credit packs start at $0.015 per credit for 250 credits and drop by $0.0001 per credit for each additional 50 credits, reaching $0.0105 at 2,500; pack credits do not expire. The monthly plan starts at $0.01 per credit and reaches $0.0055 at 2,500; its credits arrive every month, are spent first, and expire at the end of the month. Cancel it any time. The selected amount and total are shown before checkout. The desktop app is a separate $19 one-time purchase with Standard conversion; Pro recognition and translation are on the website only. Prefer not to sign in? The first 10 pages of recovered text or OCR, combined, convert free with no account at all. A book that needs more is converted with credits, so it needs an account, which starts with 30 free credits. Pro recognition always requires an account and credits.

Standard converts the PDF's own text with a rule-based engine, locally and free for text PDFs. It finds chapters, strips headers and footers and keeps images, and runs OCR on scanned pages when needed. It does well on novels, reports and other simple layouts.

Pro is AI conversion. An AI document-recognition model reads every page as an image, the way a person would, and rebuilds reading order across columns, headings, tables, formulas, code and images. That gives a cleaner book from any PDF, not only textbooks: complex layouts, scans and PDFs with a broken text layer benefit most. Pro uses 2 credits per PDF page because every page is read by the AI model.

Any ordinary text PDF can be converted with Standard for free, without an account or page limit. Automatic chapter detection, removal of repeated headers and footers, image preservation, a clickable table of contents, and EPUB formatting for your selected reader are included.

Fixed Layout is also free because it preserves each PDF page as an image. Credits are only required when Standard must recover text from scanned pages or when you choose Pro recognition.

DOCX, RTF, TXT, ODT and HTML files convert free too, with no account and no page limit — they already carry real paragraph and heading structure, so there's no scan or OCR question to settle.

It was almost certainly a scan. If it holds a recognised text layer this converter will find and use it. If it holds none, tick "Read scanned pages" to run OCR, or accept a fixed-layout book of page images.

Turn it on for scanned pages that contain no selectable text. The quote checks the PDF first: existing text pages remain free, and so do the first 10 pages that need OCR; past those, each uses 1 credit. If you leave OCR off, those pages are preserved as images instead and do not use credits.

Yes. Amazon's Send to Kindle accepts EPUB, by email or through their web uploader. Choose your reader under Target device first and the book is sized for it.

Anonymous uploads and EPUBs are removed after download or shortly afterwards.

If you sign in, uploaded files and completed EPUBs stay in your library for repeat downloads for up to 30 days; they are not permanent cloud storage. Reserved credits are returned automatically when a conversion fails.

100 megabytes per file. A long book is fine; conversion runs in the background and the page shows which stage it has reached.