Extract a page range into a new PDF, or split every page into its own file. Click pages in the thumbnail grid or type a range like 1-3, 5, 8-10 — nothing is uploaded, everything happens on your device.
Drop a PDF here or browse
Extract a page range, or split every page into its own file
🔒 Your PDF is split entirely on your device — nothing is ever uploaded.
Your document stays on your device — unlike splitters that upload your PDF to a server.
See every page as a thumbnail and click to add or remove exactly the ones you need.
Extract one range into a single PDF, or split every page into its own file inside a ZIP.
This page uses two separate PDF engines because it needs two different things. PDF.js — Mozilla's renderer — draws each page as a thumbnail so you can see what you are selecting; each one is rendered to a small canvas about 150 pixels wide. pdf-lib does the actual extraction, building a new document and copying your chosen pages into it.
The distinction matters because the thumbnails are only a preview. The pages that end up in your output are copied structurally — content streams, embedded fonts, images and vector artwork carried across intact — not re-rendered from those pictures. Text in the extracted file stays selectable and searchable, and the pages are pixel-for-pixel identical to the original at any zoom level. If thumbnail rendering fails for some reason, the tool hides the grid and the text range field carries on working, because extraction never depended on it.
Selection works both ways. Clicking a thumbnail toggles that page and rewrites the range field, collapsing runs of consecutive pages into ranges. Typing in the field re-highlights the matching thumbnails. Use whichever suits the document — thumbnails when you are looking for something visually, the text field when you already know the numbers.
| You type | You get | Note |
|---|---|---|
1-3, 5, 8-10 | Pages 1, 2, 3, 5, 8, 9, 10 | Ranges and single numbers can be mixed freely. |
5, 1-3 | Page 5 first, then 1, 2, 3 | Pages come out in the order you typed them — this is how you reorder while extracting. |
8-5 | Pages 5, 6, 7, 8 | A backwards range is corrected rather than reversed. There is no way to reverse pages within a range. |
2, 2, 3 | Pages 2, 3 | Duplicates are removed. A page cannot appear twice in one extraction. |
1-3, 99 | Pages 1, 2, 3 | Numbers beyond the last page are silently ignored. |
The ordering behaviour is the one worth remembering. Clicking thumbnails always produces a sorted list, but typing does not — so if you need the appendix to come before the introduction, type the ranges in that order and extract in a single pass instead of extracting twice and merging.
Split every page can produce more data than you started with. Each single-page PDF is a complete, self-contained document, so it embeds its own copy of every font used on that page. A 200-page report that embedded one typeface once now embeds it 200 times, and the ZIP can easily exceed the original file. It is also slow: each page means building and saving a separate document. For a long document, extracting the ranges you actually want is faster and far smaller than splitting everything.
Removing pages is not redaction. Extracting pages 1–5 genuinely drops pages 6 onwards — those pages and their content are gone from the new file. But anything sensitive on the pages you kept is still there in full, including text sitting underneath a black rectangle someone drew in a viewer. That kind of covering is a graphic on top of the text, not a removal of it, and it can be selected and copied straight out of the file. If a document needs to be safe to publish, the text itself has to be removed, not covered.
Document-level features do not survive. Bookmarks, the outline tree, the accessibility structure used by screen readers and the original document's metadata all belong to the document rather than to its pages, so the extracted file starts without them. Interactive form fields are unreliable for the same reason. Any digital signature is void — a signature certifies an exact set of bytes, and this is a new file.
Encrypted PDFs cannot be read. A file that needs a password to open reports an error rather than producing a broken result. Open it in a PDF viewer, enter the password, save an unprotected copy, and work from that. The whole document is also held in memory while you work, and a very long file means rendering a thumbnail for every page — expect the grid to take a moment on a few hundred pages.
The most common one is trimming before sending. A 90-page contract where the other party needs three pages should be sent as three pages — smaller, faster to review, and it does not hand over material nobody asked for. Extract, check the result opens correctly, then attach that.
Splitting also makes rearranging possible. To move a section, extract the pieces in the order you want using the typed-order behaviour above. To insert something into the middle of a document, split the original at the insertion point and combine the parts with the new material in Merge PDF — the working file passes straight between the two tools, so there is no saving and re-opening in between. And to interleave a double-sided scan where odd and even pages landed in separate files, split both into single pages first, then merge them alternately.
Split every page earns its place in a narrower set of cases: when each page is genuinely a separate document — a batch of scanned invoices, one certificate per page, statements to be filed individually. The ZIP names files with zero-padded page numbers, so they sort correctly in a file manager rather than putting page 10 before page 2.
If the extracted file is still too large to send, the PDF compressor reduces it further, and PDF to JPG converts pages to images when the recipient needs a picture rather than a document. The full set is on the PDF tools page.
Yes — completely free, no sign-up, no watermark added by us, and no page limits.
No — splitting happens entirely in your browser using pdf-lib and PDF.js, so your PDF never leaves your device.
Type page numbers and ranges separated by commas, e.g. 1-3, 5, 8-10. You can also click pages in the thumbnail grid to add or remove them — the text field updates automatically. Leave it empty to keep every page. Numbers beyond the last page are ignored, duplicates are removed, and a backwards range like 8-5 is read as 5-8.
One single-page PDF for every page in your document, all downloaded together in a single ZIP file, numbered in order. Be aware that each file is self-contained and re-embeds the fonts its page uses, so on a long document the ZIP can end up larger than the original PDF.
Yes — pages are extracted in the order you type them. Entering 5, 1-3 gives you page 5 first, followed by pages 1 to 3. Clicking thumbnails always produces a sorted selection, so use the text field when the order matters. Reversing pages within a single range is not possible; list them individually instead.
Only for the pages you actually removed — those are genuinely gone. Anything on the pages you kept remains fully present, including text hidden beneath a black box drawn in a PDF viewer, which is a graphic laid over the text rather than a removal of it and can be selected and copied out. For a document that must be safe to publish, the sensitive text itself has to be deleted, not covered.