A scanned book is a single 400-page file, and nobody wants a 400-page file: not the reviewer who needs chapter six, not the e-reader that chokes on it, not the portal with a 10 MB limit. Splitting it well depends on one question — does the document know where its chapters are? — and the answer decides which of the methods below to use.
First, find out what the PDF knows
Open the file in any viewer and check two things:
- Does it have bookmarks (an outline in the side panel)? A publisher’s PDF usually does; a scan almost never does.
- Can you select text? If so, it has a text layer — either born-digital or OCR’d. If dragging across a paragraph selects nothing, it is a bare scan.
Bookmarks give you the easiest split. A text layer gives you a good one. A bare scan gives you the manual options, or a quick trip through OCR PDF to earn the better ones.
The methods
Split PDF and Split PDF by size or bookmark between them cover every case. All of them run in your browser; the book is never uploaded.
By bookmarks
Each top-level bookmark starts a new part, named after it. The cleanest result when the outline is real. The tool explains the rule before it runs and warns when bookmarks do not point at usable pages.
By page-start words
Give it a word or phrase that a chapter’s opening page carries — Chapter, a case name — and every page whose first 900 characters of text contain it, ignoring capitals, opens a new part. It is a plain text match, so it counts wherever the letters fall in that stretch: a running head that repeats the word at the top of every page, a sentence that mentions a chapter, or part inside party all start a new part. Choose a phrase only a chapter opening has. Needs a text layer. Suits OCR’d scans, where bookmarks never existed but the printed headings did; a heading the OCR misread will not match.
By blank separator sheets
Photocopy pilers and scanning services insert a blank page between documents. Filozy renders each page at low resolution, treats any page with almost no ink as a separator, splits there and leaves those pages out of every file. Works on bare scans with no text layer at all. A chapter’s last page with only a line or two and a page number can count as blank and disappear, so check that the number of separator pages in the result headline matches the sheets you inserted.
By custom ranges
Type the page ranges yourself — 1-12, 13-40, 41-88. The fallback that always works, and the right choice when you have a table of contents in front of you.
Every N pages
For reading in sittings: a 310-page book at every 20 pages becomes 16 files, the last holding the 10 pages left over. A stack of scanned two-page forms becomes one file per form with every 2 pages. Arithmetic, no inspection of the content, so it works on a bare scan too.
Into N equal parts
Hand a manuscript to four reviewers. Also arithmetic.
By target file size
For upload limits: parts that each stay under a size you choose, measured from the actual output. See the caveats in the FAQ — sizes are approximate by nature, and a single oversized page cannot be split further.
Step by step for a scanned book
- Trim the margins with Auto crop if you will read the parts on a phone or e-reader: the text fills more of the screen. It changes only the visible page area, so it does not straighten a crooked scan or make the file smaller.
- OCR it with OCR PDF if you want chapter detection by words, searchability in the parts, or both.
- Open Split PDF by size or bookmark. Choose At bookmarks if the file has an outline, At page-start text and the chapter word if it has a text layer, At blank pages for a bare scan with separator sheets, or Every N pages for parts of equal length. To type ranges from the table of contents instead, use Split PDF.
- Read the result. The headline says how many files were made, and the notes under it say how each cut was made. If a cut is wrong, change the word or the method and split again.
- To keep only some chapters in one file, use Extract PDF pages on the original instead.
- Save. Parts download individually or as a ZIP.
- Open two or three parts and check the first and last pages of each. Keep the original.
Naming and order
Bookmark splits are numbered in document order and named after the bookmark, such as book-06-Chapter-Six.pdf in a book with ten or more parts, so they sort in reading order. Every other method numbers the parts after the source name (book-1.pdf, book-2.pdf … book-16.pdf), which a file manager that sorts numbers naturally lists in order, but a plain alphabetical sort puts book-10 straight after book-1. If you rename them, put a zero-padded number first; 06 Chapter Six.pdf survives any file manager, Chapter Six.pdf ends up between Chapter Seven and Chapter Ten.