Split a PDF by its bookmarks
Splitting a PDF by bookmarks reads the document's outline, resolves each top-level bookmark to the page it points at, and cuts the file at those pages — producing one PDF per chapter, automatically named after the bookmark titles. Pages before the first bookmark are saved as a front-matter part so nothing is dropped. This page performs the whole split in your browser; the file is never uploaded.
How it works
Bookmarks live in the PDF's outline dictionary: a tree of entries, each holding a title string and a destination. The tool walks the first level of that tree and resolves every entry's destination to a concrete page index. That resolution is the fiddly part — a destination can be an explicit page reference, a named destination that must be looked up in the document's name tree, or an action wrapping either — and the tool handles all three forms so that each top-level bookmark ends up mapped to exactly one page number.
Those page numbers become cut points. Chapter k spans from its bookmark's page up to the page before bookmark k+1 (the last chapter runs to the end of the document). If any pages precede the first bookmark — cover, imprint, table of contents — they are collected into a front-matter part rather than silently dropped. Each part is then built by structurally copying its page range into a fresh document, so fonts, images, and vector content are preserved bit-for-bit; nothing is re-rendered.
Output naming comes straight from the outline: each part's file name is its bookmark title, sanitized for the file system and prefixed with a sequence number so the parts sort in reading order. Only top-level entries define cuts; nested sub-bookmarks remain within their parent chapter's part. A document with no outline at all cannot be split this way — the tool reports it and points you to bookmark PDF to add an outline, or to split PDF to cut by page ranges instead. If you got the file from our merge with contents page tool, its per-file bookmarks make this split an exact reversal of the merge.
Illustrative example: a 240-page manual into 9 chapter files
Illustrative scenario, not a recorded benchmark. File sizes, timings and results depend on the input and device. Try the reproducible samples.
A technical writer has a 240-page, 18 MB product manual with nine top-level bookmarks: Introduction (page 5), Installation (page 21), Configuration (page 48), and so on through Appendix (page 214). Pages 1–4 are the cover and contents, before any bookmark.
The split produces ten files: 00 Front matter.pdf (pages 1–4, 0.3 MB), 01 Introduction.pdf (pages 5–20, 1.1 MB), 02 Installation.pdf (pages 21–47, 2.0 MB), through 09 Appendix.pdf (pages 214–240, 1.8 MB). The sizes add up to slightly more than the original 18 MB because fonts used across chapters are copied into each part that needs them. The whole run takes about eight seconds in the browser.
Frequently asked questions
What happens if my PDF has no bookmarks at all?
The tool reports an error, because there are no outline entries to cut at. You can add an outline first with our bookmark PDF tool, or split by page ranges with the regular split tool instead.
How are the output files named?
Each part is named after its bookmark title, with characters that are invalid in file names replaced. A numeric prefix keeps the parts in reading order when sorted in a folder.
What happens to pages before the first bookmark?
They are saved as a separate front-matter part rather than discarded. Title pages, tables of contents, and prefaces commonly live there, so nothing is lost.
Does the split use nested bookmarks or only top-level ones?
Only top-level bookmarks define cut points; nested sub-bookmarks stay inside their parent's part. This keeps a chapter with twelve subsections as one file instead of twelve fragments.
What if two bookmarks point at the same page?
A chapter must span at least one page, so when two top-level bookmarks resolve to the same page they are merged into a single part covering that range. The part takes the first bookmark's title.
Is the PDF uploaded anywhere during the split?
No. The outline is parsed and the parts are assembled by JavaScript in your browser, and the downloads are generated locally. The file never leaves your device.