Split PDF
Say which pages you want, like 1-3, 7, 9-12. Each range comes back as its own PDF, zipped together. All of it happens on your device.
Files never leave this device · works offline
Drop Split PDF files here
or drop them anywhere here · or paste from clipboard
No upload caps · No sign-up · No watermarks
Processing locally · 0 bytes sentHow to verify this →
Each range becomes its own file
This is the part most tools get wrong in a way that costs you time. Ask for 1-3, 7-9 and you should get two documents, because you named two things. Some tools hand back a single six-page file instead, and you are left splitting it again by hand.
Here every range you type produces its own PDF, named for the pages it holds, and they arrive together in one ZIP. If you actually want one file containing several separate stretches, say it as a single range or merge the results afterwards.
It tells you when it did not understand you
Page selections are typed by hand, so they contain typos. Something like 1-3, foo, 7 is a real thing to type. A tool that quietly ignores the part it could not read hands you a document that is wrong in a way you cannot see, and you find out weeks later when a page is missing.
Anything unreadable is reported back by name, and the ranges that did parse are still extracted. Page 0 is rejected rather than nudged to page 1, because a zero usually means a misunderstanding rather than a slip, and silently widening a selection is its own kind of wrong.
Text stays text
Pages are copied rather than re-rendered, so the extracted files keep selectable text, working links, annotations and form fields. Nothing is flattened into a picture of a page. If you split a contract to send one clause on, whoever receives it can still search it.
How ranges are read
| You type | You get |
|---|---|
| 1-3 | One file containing pages 1, 2 and 3 |
| 1-3, 7 | Two files. Pages 1 to 3, and page 7 on its own |
| 1-3, 4-8, 9-12 | Three files, one per range |
| 9- | One file from page 9 to the end |
| 7-3 | Pages 3 to 7. A backwards range is read the sensible way |
| 1-500 on a 12-page file | All 12 pages. A range past the end is trimmed, not refused |
What people use this for
- Pulling one signed page out of a long agreement to send on
- Separating a bank statement into the months a form actually asked for
- Breaking a scanned report into chapters
- Extracting just the pages a portal will accept when it caps the page count
- Removing pages you would rather not share, without opening an editor
Frequently asked questions
Are my PDFs uploaded anywhere?
No. Everything runs in your browser on your own device, and nothing is sent over the network at any point. Open the Network tab in developer tools and split a file. You will see no request carrying it, because there is none. Disconnecting from the internet and splitting anyway is the quicker proof.
How do I write the page ranges?
Commas separate ranges and a dash makes a span, so 1-3, 7, 9-12 gives you three files. You can leave one side of a dash open: 9- means page 9 to the end, and -3 means the start through page 3. Backwards ranges like 7-3 are read as 3 to 7, and spaces make no difference.
Why do I get a ZIP instead of separate downloads?
Because a browser asking permission for eleven separate downloads is worse than one archive. Each PDF inside is named for the pages it contains, so you can see what is what before extracting anything.
Will the extracted pages still have selectable text?
Yes. Pages are copied as pages, so text stays text and links keep working. Nothing is rasterised into an image, which is what happens if a tool prints to PDF rather than genuinely extracting.
What if I ask for pages that do not exist?
A range running past the end is trimmed to the last page, since 1-500 on a twelve-page file plainly means all of it. A range that starts past the end is reported as unusable rather than silently ignored, so you know it did not do what you asked.
Does it work offline?
Yes, after the first visit. There is no server involved, so there is nothing to be offline from.
Nothing is uploaded — and you can check
Conversion happens on your own device, inside this browser tab. There is no server to receive your files, which is a stronger guarantee than a promise not to look at them. Open your browser's developer tools, switch to the Network tab, and convert something: you will see no request carrying your image. The page keeps working with your network disconnected.