PDF Splitter

Split large PDFs into smaller, focused documents before importing them into NotebookLM.

Split PDFs by page range, extract chapters, or break documents into equal parts — entirely in your browser. Built for NotebookLM users. No uploads, no sign-up, files never leave your device.

NotebookLM works best when every source is focused. Upload a 600-page textbook and you get a notebook that has to sift through fourteen chapters of context for every question — summaries drift, citations land in the wrong chapter, and the one appendix you actually care about gets diluted by everything around it. Most of the time you don't need the whole book: you need Chapter 3, pages 40–72, the appendix, or one paper out of a conference proceedings file. This PDF splitter makes that extraction take seconds. Drop in a PDF and pull out exactly the pages you need — a single range, several ranges at once, equal parts, hand-picked pages, or every page as its own file. Splitting happens entirely on your device in a background thread, pages are copied without re-rendering so formatting is untouched, and the output files get names that still make sense next week — research-paper-pages-20-35.pdf, not document(2).pdf.

How it works

1. Upload a PDF — Drag and drop or browse for your file. You'll immediately see the page count, file size, metadata, and a browsable preview of every page — click any page to enlarge it and find your split points. Nothing is uploaded anywhere.
2. Choose how to split — Extract one page range, several ranges at once, equal parts of N pages, hand-picked individual pages, or every page as its own file.
3. Download focused PDFs — Files come out with meaningful names — like research-paper-pages-20-35.pdf — ready to import into NotebookLM. Multiple files can be grabbed one by one or as a single ZIP.

Why this helps NotebookLM

NotebookLM builds its answers from the sources you provide, and source quality is mostly about focus. A source that contains one chapter, one paper, or one appendix gives the model a clean boundary: everything inside is relevant, everything outside doesn't exist. A source that contains an entire book forces the model to guess which of its 600 pages you actually meant.

Focus also shows up in citations. When your source is pages 103–155 of a textbook, every citation NotebookLM gives you lands inside the material you're studying — you can click through and verify it in seconds. When your source is the whole book, citations scatter across chapters you never intended to include, and verifying them means paging through material you were trying to avoid.

And there's a practical dividend: focused sources make focused notebooks. One notebook per topic, each fed with just the relevant chapters and papers, beats a single mega-notebook holding everything you've ever read. Splitting PDFs before import is the unglamorous step that makes that structure possible.

What you need to know

Why focused sources beat whole books in NotebookLM

NotebookLM grounds every answer in the sources you give it, and it treats each source as one unit of context. When that unit is an entire textbook, the model has to weigh material from every chapter every time you ask a question — so a question about photosynthesis can surface citations from the ecology chapter, and a study guide ends up skimming fourteen topics instead of mastering one. A source that contains only the chapter you're studying changes that completely: summaries stay on topic, citations point where you expect, and audio overviews follow one argument instead of hopping across a whole book. The habit that separates organized notebooks from chaotic ones is simple — import the section you're actually researching, not the file it happens to live in.

How to choose your split points

The table of contents is your map. Open the PDF in any viewer, note where the section you need starts and ends, and enter that as a page range — 103-155 for Chapter 3, say. Watch for one common trap: printed page numbers rarely match PDF page numbers, because front matter (title pages, contents, prefaces) shifts everything by ten or twenty pages. Check the actual page position in your PDF viewer's page counter, not the number printed on the page. For research papers bundled into proceedings, split at each paper's first page. For reports, separate the executive summary, the body, and the appendices — they serve different questions, and they work better as different sources.

Splitting also keeps you inside NotebookLM's limits

NotebookLM enforces per-source limits — at the time of writing, a single source can hold up to 500,000 words, and uploads are capped at 200 MB. Long technical books, scanned archives, and multi-volume PDFs can hit those ceilings, and when they do, the import simply fails or truncates. Splitting solves this mechanically: four 150-page parts import cleanly where one 600-page file might not. It also spends your source slots more wisely. Each notebook has a limited number of sources, so it's tempting to cram everything into one giant file — but three focused documents that NotebookLM can cite precisely will outperform one monolith that it can only cite vaguely.

Your files never leave your browser

This tool has no upload step, no server-side processing, and no AI. The PDF is read directly by JavaScript running on your device, pages are copied into new documents by an open-source PDF library working in a background thread, and the results are handed straight to your browser's download manager. Nothing is transmitted, stored, or logged — you can open this page, disconnect from the internet, and split files all day. That matters more for PDFs than for most file types, because the documents people split are often the sensitive ones: legal case files, unpublished manuscripts, internal reports, medical research. Client-side processing isn't a feature toggle here; it's the architecture.

Common use cases

• Textbook chapters for study notebooks — Pull Chapter 3 out of a 600-page textbook and build a notebook around just that exam topic. Study guides and quizzes stay on the material you're actually being tested on.
• One paper from conference proceedings — Proceedings and anthologies bundle dozens of papers into one PDF. Extract the single paper you're citing so NotebookLM treats it as its own source with its own citations.
• Case files and contracts for legal research — Separate exhibits, opinions, and individual agreements from a large filing — locally, so sensitive documents never touch a server — and query each one precisely.
• Report sections for briefings — Split an annual report or market study into executive summary, body, and appendices. Build the briefing notebook from the sections that matter and leave the boilerplate out.

What’s preserved

• Page content — text, images, vector graphics
• Fonts and layout, exactly as in the source
• Selectable text layers (what NotebookLM reads)
• Page size and orientation
• Page order within your selection

What’s removed

• Pages outside your selection
• Hundreds of pages of noise your notebook doesn't need

Frequently asked questions

Does my PDF leave my computer?

No. The file is read and split entirely by JavaScript running in your browser — there is no upload, no server-side processing, and no AI involved. Nothing is transmitted, stored, or logged. The heavy work runs in a background thread on your own device, which is also why there's no file-size pricing tier: there's no server bill to pass on.

Can I split password-protected PDFs?

Not directly — encrypted PDFs can't be split without decrypting them first, and this tool never asks for your password. Remove the protection first: open the PDF in a viewer where you know the password, unlock it, and use Print → Save as PDF to produce an unprotected copy. Then split that copy here.

Can I extract only one chapter?

Yes — that's the main use case. Find the chapter's start and end pages in the table of contents, check them against your PDF viewer's page counter (printed page numbers usually differ from PDF page positions because of front matter), and enter the range — for example 103-155. You'll get a single PDF named after your document and that range.

Does this work offline?

Yes. Once the page has loaded, no network connection is needed — you can disconnect entirely and keep splitting files. Everything the tool needs runs locally in your browser.

Can I split very large PDFs?

Yes. Processing runs in a background thread, so the page stays responsive even with documents that have hundreds or thousands of pages. The practical ceiling is your device's memory rather than any limit we impose — files of several hundred megabytes work on a typical laptop. If a huge scanned file feels slow, splitting it in a few passes (halves first, then chapters) keeps memory use down.

Will formatting change?

No. Pages are copied from the original document, not re-rendered or re-compressed — fonts, images, vector graphics, layout, page size, and selectable text layers come through exactly as they are in the source file. The output is the same PDF content, just less of it.

Why split a PDF before importing it into NotebookLM?

Two reasons. Quality: NotebookLM's summaries, study guides, and citations are noticeably sharper when a source contains only the material you're researching, instead of a whole book it has to sift through. Limits: NotebookLM caps each source (500,000 words and 200 MB per upload at the time of writing), and splitting is the reliable way to get long documents under those ceilings.

← Back to Toolkit · Resources