How to download a book from the Internet Archive
- Open the book's page on archive.org. Public-domain scans can be downloaded; borrowable books cannot.
- Find the Download options box on the right side of the page.
- Pick PDF for the scan. The list may also show EPUB, FULL TEXT and DjVu files.
- If you only see a "Borrow" button, the book is in controlled lending and there is no file to download.
The Archive's help pages say that EPUB, DjVu and the other derived formats are made from OCR, and that OCR "can be poor". Those words are worth taking literally. The EPUB is a by-product of the search index, not an edition someone checked.
What is inside the Archive's EPUB: one book, measured
We downloaded the Archive's EPUB of Alice's Adventures in Wonderland (Macmillan, 1872) and compared it word by word with the proofread Project Gutenberg text. Here is what we found.
| Check | Internet Archive EPUB | upub EPUB |
|---|---|---|
| Table of contents entries | 0 | 39 |
| Book title in the metadata | missing | Alice's Adventures in Wonderland |
| Paragraphs | 207 (one per page) | 906 |
| Pages marked "estimated to be only X% accurate" | 34 | 0 |
| Words in the story | 29,173 | 26,689 |
| EPUBCheck errors | 32 | 0 |
The word count tells the story. The proofread text has 26,681 words. The Archive's EPUB has 29,173, because every printed running head and page number is still in the text. A typical page opens with "4 DOWN THE" in the middle of a sentence, and OCR slips such as "w^onder" and "merr}' crew" are left as they are.
| Method | Words recovered | Clean text | Contents entries | EPUBCheck | File size |
|---|---|---|---|---|---|
| Internet Archive's EPUB | 96.8% | 88.5% | 0 | 32 errors | 12.1 MB |
| Calibre on the scan | 0% | – | 0 | valid | 44.0 MB |
| OCRmyPDF, then Calibre | 0% | – | 0 | valid | 43.9 MB |
| upub | 99.2% | 99.2% | 39 | valid | 167 KB |
Measured on 2026-10-02 against the proofread Project Gutenberg #11 text (26,681 words). Words recovered: the share of that text found in the EPUB, in order. Clean text: the share of the EPUB's words that belong to the book, not to running heads, page numbers or OCR noise.
Where the Archive's file is better: it keeps the 47 Tenniel illustrations as images. Our sample carries the text only, because upub cannot yet cut figures out of a full-page scan. If the pictures matter more to you than the text, keep both files.
Why the free EPUB looks like this
OCR reads each page as an image and returns lines of text. It does not know that the line at the top is a running head, that a hyphen at the end of a line splits one word, or that "CHAPTER II" starts a new section. An e-book needs all three decisions. When nobody makes them, you get page-shaped paragraphs and a table of contents with nothing in it.
upub uses an AI model that sees the whole page, so it can tell a heading from a running head and a footnote from body text. Then it joins paragraphs across page breaks and writes an EPUB 3 with a real chapter menu.
Google Books, HathiTrust and other scan libraries
Google Books lets you download full-view books as PDF or EPUB from the book's page. The same rules apply as for the Archive: the EPUB is generated from OCR, and the PDF is a scan. Whether a Google PDF also carries a hidden text layer varies from book to book, so check before you rely on it.
- Check for text: open the PDF and try to select a sentence. If the whole page highlights as one block, it is images only.
- HathiTrust offers public-domain volumes page by page or as a full PDF when you are signed in through a member library.
- Projekt Gutenberg and Project Gutenberg already publish proofread EPUBs. If your book is there, you do not need to convert anything.
Turning an Archive PDF into a clean EPUB
- Download the PDF from the Download options box, not the EPUB.
- Upload it to upub and pick your reader, for example Kindle Paperwhite or Kobo Libra Colour.
- upub recognises Internet Archive PDFs and reads the page images again instead of copying the Archive's OCR layer.
- Download the EPUB and send it to your device. For Kindle, our Kindle guide covers Send to Kindle.
Questions
- Is the Internet Archive's EPUB good enough to read?
- Sometimes. It is generated from raw OCR, so it often keeps running heads and page numbers in the text and has an empty table of contents. On our 1872 test book it failed EPUBCheck with 32 errors.
- Which file should I download from archive.org to convert?
- The PDF. It is the scan itself, so a converter can read it again from scratch instead of inheriting the Archive's OCR mistakes.
- Can I convert a borrowed Internet Archive book?
- No. Books in controlled lending are protected and cannot be downloaded as files. Only public-domain and openly licensed items have a PDF in the Download options box.
- Does Google Books give a PDF or an EPUB?
- For full-view books Google's help page offers both. The PDF is a scan; the EPUB is generated from OCR.
Try it on your own scan
Upload a scanned PDF or a folder of page photos. You get an EPUB with real text, chapters and a table of contents.
Sources
- Internet Archive Help: Books and texts tips and troubleshooting
- Internet Archive Help: Books and texts, a basic guide
- Internet Archive blog: Upcoming changes in EPUB generation (2016)
- Google Books Help: Download books
- Test scan: Alice's Adventures in Wonderland, Macmillan 1872
- Reference text: Project Gutenberg #11