Will I be able to select and search the text?

Yes. The PDF contains real text, so search, copy and select all work, and a screen reader can read it. That is the main thing separating this from converters that photograph a reader — those produce a PDF of pictures with no text in it at all.

Which languages can it typeset?

The Latin alphabet including every accented European language, plus Greek and Cyrillic — so Russian, Ukrainian, Bulgarian, Serbian and Greek all set properly. Latin books need no download; a book with Greek or Cyrillic in it fetches a proper book face (about 900 KB, once) so nothing is dropped. It still cannot set Chinese, Japanese, Korean, Hebrew, Arabic or Devanagari, and it tells you that before you convert rather than handing you a file full of blank boxes.

Is there a model? How big is the download?

There is no model and there is no download. This is typesetting, not machine learning — the whole thing is a few hundred kilobytes of code that is already on the page by the time you drop your file.

What happens to the cover and the table of contents?

Illustrations inside the book are kept and placed on the page. The EPUB's own navigation list is left out, because a table of contents with no page numbers in it is not useful in a PDF — the chapters themselves start on fresh pages, which is what you actually navigate by.

What happens to tables and footnotes?

Both are typeset properly rather than flattened. A table becomes a real grid: columns sized to what is in them, numeric columns ranged right, and the header row repeated if the table runs past a page break. A footnote is placed under a short rule at the foot of the page its marker falls on — not gathered into endnotes — and a long note carries over to the next page rather than overrunning the margin. One caveat: a contents page printed inside the book is kept as the author set it, page numbers and all, and those are the print edition’s numbers, not this PDF’s.

Why are there four styles instead of one?

Because people want different things from a book file. Classic is a novel: Times, justified, indented paragraphs, a drop cap. Modern reads like a report — sans-serif, ragged right, spaced paragraphs. Manuscript is Courier and double-spaced, the format editors and translators mark up by hand. Large print is 16 pt with generous leading.

Can it do fixed-layout EPUBs, like comics or picture books?

No. A fixed-layout EPUB is already a set of designed pages, usually one image each, and reflowing it makes no sense. This tool is for books that are mostly text.

Are my files uploaded?

Your files are never uploaded. The EPUB is read and the PDF is built inside your browser tab, which is also why it works with no network and why there is no size limit imposed by a server.

How long does a big book take?

A 400-page novel takes about a second. The Pickwick Papers — 1.7 million characters, 970 pages — takes around two and a half. There is no queue and no upload, so the time you see is the actual work.

Turn an ebook into a PDF you can print

An EPUB reflows to fit whatever screen it is on. A PDF has pages. Turning one into the other means deciding where every line and every page breaks — and most converters dodge that by photographing a reader, which is why their PDFs are enormous and you cannot search or select a word. This one typesets the book properly, so what comes back is real text on real pages.

Your EPUB
Drop your EPUB here
EPUB 2EPUB 3
Your files are never uploaded.
How it should look
Your PDF

How to convert an EPUB

Drop in your EPUB

It is read in the browser. The file never leaves your computer, so there is nothing to upload and nothing to wait for.
Pick how it should look

Four page styles, shown as miniatures rather than named — Classic reads like a novel, Manuscript is the double-spaced format editors mark up. Choose a page size while you are there.
Get your PDF

Text you can select, search and copy, with the illustrations in place. A 400-page novel comes out around a megabyte and takes a couple of seconds.

Why this one looks like a book

Real type, not pictures of type

Most converters screenshot each screen of an ebook reader. You get a PDF of images: nothing is searchable, nothing is selectable, and the file is many times larger than it needs to be. Here the words are words.

It hyphenates, like a book does

Justified text without hyphenation leaves gaping word gaps — about one line in five. The Classic style uses the same algorithm TeX does, in the language of the book, which cuts that to roughly one in twenty.

Chapters start on a new page

Chapters are found by their headings, not by how the EPUB happens to split its files, so a chapter opening is a chapter opening — running head, folio, and a drop cap in the Classic style.

Tables stay tables, footnotes stay at the foot

A table is set as a real grid — columns sized to their content, numbers ranged right, the header repeated if it runs over a page. Footnotes go under a short rule at the foot of the page their marker is on.

The illustrations come with it

Plates and line drawings are placed at their own size, centred, with captions set beneath them. An illustrated Alice keeps all of Tenniel.

Large print is a real option

16 pt Times, loose leading, high contrast — and no hyphenation, because words split across lines are harder to read for exactly the people who need this.

No model, and no upload

There is no machine learning here at all, and your book is never sent anywhere. A Latin-alphabet book downloads nothing; one with Greek or Cyrillic fetches a book face once, and that is a font, not a model.

Questions about EPUB and PDF

Will I be able to select and search the text?: Yes. The PDF contains real text, so search, copy and select all work, and a screen reader can read it. That is the main thing separating this from converters that photograph a reader — those produce a PDF of pictures with no text in it at all.
Which languages can it typeset?: The Latin alphabet including every accented European language, plus Greek and Cyrillic — so Russian, Ukrainian, Bulgarian, Serbian and Greek all set properly. Latin books need no download; a book with Greek or Cyrillic in it fetches a proper book face (about 900 KB, once) so nothing is dropped. It still cannot set Chinese, Japanese, Korean, Hebrew, Arabic or Devanagari, and it tells you that before you convert rather than handing you a file full of blank boxes.
Is there a model? How big is the download?: There is no model and there is no download. This is typesetting, not machine learning — the whole thing is a few hundred kilobytes of code that is already on the page by the time you drop your file.
What happens to the cover and the table of contents?: Illustrations inside the book are kept and placed on the page. The EPUB's own navigation list is left out, because a table of contents with no page numbers in it is not useful in a PDF — the chapters themselves start on fresh pages, which is what you actually navigate by.
What happens to tables and footnotes?: Both are typeset properly rather than flattened. A table becomes a real grid: columns sized to what is in them, numeric columns ranged right, and the header row repeated if the table runs past a page break. A footnote is placed under a short rule at the foot of the page its marker falls on — not gathered into endnotes — and a long note carries over to the next page rather than overrunning the margin. One caveat: a contents page printed inside the book is kept as the author set it, page numbers and all, and those are the print edition’s numbers, not this PDF’s.
Why are there four styles instead of one?: Because people want different things from a book file. Classic is a novel: Times, justified, indented paragraphs, a drop cap. Modern reads like a report — sans-serif, ragged right, spaced paragraphs. Manuscript is Courier and double-spaced, the format editors and translators mark up by hand. Large print is 16 pt with generous leading.
Can it do fixed-layout EPUBs, like comics or picture books?: No. A fixed-layout EPUB is already a set of designed pages, usually one image each, and reflowing it makes no sense. This tool is for books that are mostly text.
Are my files uploaded?: Your files are never uploaded. The EPUB is read and the PDF is built inside your browser tab, which is also why it works with no network and why there is no size limit imposed by a server.
How long does a big book take?: A 400-page novel takes about a second. The Pickwick Papers — 1.7 million characters, 970 pages — takes around two and a half. There is no queue and no upload, so the time you see is the actual work.

Not sure this is the conversion you want? EPUB vs PDF is the one difference between the formats and what converting costs, and what an EPUB file actually is explains the missing page numbers and why a DRM'd book will not open here.

Got the PDF and want it smaller before emailing it? Compress it. Printing only part of the book? Delete the pages you don't need or crop the margins to fit more on a sheet. Going the other way — a PDF you want as plain text — is PDF to Markdown.

Your EPUB

How it should look

Your PDF