Several ranges in one pass, or one combined file — your choice. By ranges: type 1-3, 7-9, 15- and get three files back, zipped together. Extract pages: type 2, 5, 8-10 and get one five-page PDF holding just those pages, in order, duplicates removed. Most splitters force one or the other; here it is a mode switch.

Nothing is uploaded and there is no limit — no daily cap, no size ceiling, no account. The file is read by this tab and the pieces are written by it.

Split PDF

Local

Choose a PDF, or drop one here Nothing leaves this tab

Written by pdf-lib, fetched when you add a file. The ZIP is assembled here — no second library.

Every expression above is accepted by both modes below. The permissive parsing is deliberate: a tool that rejects “11-9” with an error is making the visitor do work a computer can do for them. What differs between modes is what happens to the matched pages afterward — By ranges keeps each comma-separated group as its own file; Extract pages combines everything shown above into one.

Writing the ranges

The notation is the same one printers and page-range dialogs have used for decades:

Typed	Means	Result on a 20-page document
`7`	One page	A one-page file
`5-9`	A span	A five-page file
`12-`	From here to the end	Pages 12 to 20
`-4`	From the start	Pages 1 to 4
`1-3, 7-9, 15`	Three separate files	Three files: 3 pages, 3 pages, 1 page
`9-5`	Reversed span	Read as 5–9; the order is corrected rather than rejected
`odd` / `even`	Alternating pages	All odd-numbered pages, or all even-numbered ones

Page numbers here are positions in the file, not what is printed on the page. A report with a cover and two front-matter pages has its "page 1" at position 4. Check the preview above the button — it lists exactly which pages land in which file before anything is written.

Why a ZIP, and why no library for it

Splitting a twenty-page document into single pages produces twenty files. Browsers handle that badly — some block downloads after the first few, some ask for permission each time, and all of them scatter the results across a folder.

A ZIP solves it, and building one turns out not to need a dependency. A stored — uncompressed — ZIP is four small structures and a CRC-32 checksum per entry, around sixty lines of code. PDFs are already compressed internally, so deflating them again saves roughly two per cent for a noticeable delay. Storing them is both simpler and, for this content, barely larger.

Approach	Extra weight	Trade-off
Written here, stored	None	About 2% larger than a deflated archive, on content that does not deflate
A ZIP library	~95 KB	Real compression, on files that are already compressed
Separate downloads	None	Twenty prompts, twenty files, browser download limits

What each piece keeps

Element	Survives	Note
Page content	Always	Copied exactly — nothing is re-encoded, so quality and file structure are untouched
Fonts	Always	Each output embeds only the fonts its pages use, which is why the pieces usually total less than the original
External links	Always	Web addresses are annotations on the page
Internal links	Partly	A link to page 40 works only if page 40 is in the same output file
Bookmarks	No	The outline belongs to the document, not the pages
Form fields	Yes	Fields stay with their pages
Signatures	No	A signature certifies one whole file; splitting it necessarily breaks that
Encryption	No	The outputs are unprotected

Compared with the alternatives

Tool	File goes	Free limit	Several ranges at once
This page	Nowhere	None	Yes, in one pass, zipped
Smallpdf	Their servers	2 tasks a day	One operation per run
iLovePDF	Their servers	100 MB, rate limited	Ranges, one mode per run
Sejda	Their servers	50 MB, 3 tasks an hour	Ranges, one mode per run
PDFsam Basic	Nowhere	None	Yes — a desktop install

Splitting for a purpose

Most splits are done for one of a handful of reasons, and each wants a different range:

Why	Range to use	Note
Email size limit	Even chunks	Most services cap attachments around 25 MB. Splitting by page count rarely gives even sizes, because scanned pages are far heavier than text ones
Sending one section	A named range	Send only what is needed rather than the whole file with a “see page 40” note
Removing pages	Everything except them	Extract the pages you want to keep. The unwanted content is genuinely gone, not hidden
Double-sided scan	`odd` then `even`	A sheet-feed scanner run twice produces fronts then backs. Splitting separates them for reordering
Printing a section	Usually none	Print dialogues take page ranges directly. Splitting first is an extra step for nothing

Extracting pages is the safest way to remove content from a PDF. Deleting pages in an editor can leave the objects in the file for a determined reader; building a new document from only the pages you want cannot, because the discarded content was never written.

What each piece keeps and loses

Element	In the split file	Note
Page content	Identical	Text, images and vectors are copied byte for byte. No re-encoding, no quality loss
Page size and rotation	Preserved	Including mixed sizes within one document
Links to other pages	Broken	An internal link to page 40 has nowhere to go if page 40 is not in the piece
Bookmarks	Usually lost	The outline tree refers to the original page objects
Digital signatures	Invalidated	A signature covers the whole document. Any change breaks it, by design

Splitting does not shrink pages proportionally. A 100-page 50 MB file split in half rarely gives two 25 MB files — fonts and shared resources are duplicated into both, and the weight follows the images rather than the page count. Check the sizes before assuming an attachment will now fit.

Common mistakes

Using the printed page numbers. Roman numerals on the front matter, an unnumbered cover, a numbering restart at each chapter — none of that changes the position of a page in the file. Position 1 is the first page, whatever it says on it.

Splitting a document with cross-references and expecting them to work. A link to page 40 in a file that now ends at page 30 has nowhere to go. Split along the seams the document already has — chapters, sections, appendices — rather than at an arbitrary page count.

Splitting to reduce file size. The pieces total roughly what the original did — slightly less, because each embeds only the fonts it needs. If a file is too large the cause is almost always scanned pages, and splitting moves that weight around rather than removing it.

Uploading a document to split out one page. Contracts, statements and scanned identity documents are exactly what people split, and exactly what should not be posted to a server on trust. Anything local removes the question.

Frequently asked questions

Is my file uploaded?

No. It is read by your browser and the pieces are written there too — the network panel in developer tools will show no request carrying your document. The only network request is for the library, and it happens once.

Can I extract several ranges at once?

Yes, and it is the point of the tool. Type 1-3, 7-9, 15- in "By ranges" mode and you get three files in one pass, delivered as a ZIP. Most splitters run one operation at a time, so the same job takes three passes elsewhere.

What is the difference between "By ranges" and "Extract pages"?

Both read the same notation. "By ranges" keeps every comma-separated group as its own file — 1-3, 7-9 becomes two files. "Extract pages" combines everything you list into a single file instead — 2, 5, 8-10 becomes one five-page PDF holding just those pages, in order, with any duplicate pages removed.

Can I select every other page?

Yes — type odd or even in place of a range. It can be combined with other ranges too, like odd, 12.

Why is it a ZIP?

Because twenty separate downloads is worse. Browsers block or repeatedly prompt for multiple downloads and scatter the results. The archive is built here rather than with a library: a stored ZIP is a few small structures and a checksum, and PDFs are already compressed internally so deflating them again would save about two per cent.

Do page numbers mean what is printed on the page?

No — they are positions in the file. A cover and two front-matter pages put the printed “page 1” at position 4. The preview above the button lists exactly which pages go into which file, so you can check before splitting.

Will the pieces be smaller than the original?

Together, slightly — each output embeds only the fonts its own pages use. Splitting is not compression though; if the file is large the cause is usually scanned pages, and those weigh the same wherever they end up.

Do bookmarks survive?

No. The outline belongs to the document rather than the pages, so each piece starts without one. Page content, fonts, form fields and external links all come across intact.

Is there a size or page limit?

No. Server-based tools cap free use because each split costs them processing time; nothing runs on a server here. The practical ceiling is your own machine’s memory, which means hundreds of pages without trouble.

Can I split a password-protected PDF?

Not while it is protected — remove the password in a PDF reader first. Note that the outputs are never encrypted, so protect them again afterwards if the original was.

Related tools

See the full list of Document Tools, or try:

Merge PDF — the other direction, with page-level control
Compress PDF — shrink a piece further if it still does not fit a size limit
Add Page Numbers to PDF — number a piece before sending it on its own
PDF to Markdown — pull the text out
Markdown to PDF — make one from written text

Split PDF Free: Extract Pages or Split into Files