Drop a PDF into the zone below and every page appears as a tile you can see and pick. Click the pages you want, shift-click to take a whole chapter in one go, or type them as 1, 4-9, 22. Get them back as a single document or as one file per page. The pages are copied without being decoded or recompressed, the original is never modified, and nothing is uploaded to anybody.
Your files are read by JavaScript running on your own machine. There is no upload, no queue, and no copy on our server to delete later.
Drag a card to reorder it, or use the arrows under each card. The order here is the order in the output. Each card is a preview of the finished page, so what you see is what the PDF will contain.
Esc close ← → move R rotate
Starting
This is your processor doing the work, so the time it takes depends on your machine and not on how many other people are on the site.
The extractor above is ready to use. The guide below is about what you are actually handing over - why a three page extract from a 40 MB report can still be 12 MB, what an extracted page carries with it that you might not want it to, whether the pages stay in your order or the document's, and when extracting is the wrong tool and Split PDF is the right one.
Under a minute, nothing to install, no account for your first run. This PDF page extractor runs inside the page you are reading, on your own processor, and never sends your document anywhere.
Click the drop zone above or drag the file onto this page. Every page is rendered as a tile in the order it appears in the file, so you can find the pages you need by looking at them rather than by guessing at numbers.
A picked page turns emerald and is labeled Keep. Shift-click a second tile to take the whole run between them, which is how you extract a 30 page chapter in two clicks instead of thirty.
The Pages to extract field takes 1, 4-9, 22. It stays in step with the board in both directions, so you can type a range and then add one more page with a click, or start by clicking and read back the range you built.
One PDF keeps your selection together in document order. One PDF per page gives you a separate file for each sheet, named after the page it came from and delivered as a single ZIP. Then press Extract pages. The result downloads immediately.
The bar above the board carries Select all, Clear selection, Select odd pages, Select even pages and Invert. Invert is the one people underuse: it is often faster to click the four pages you do not want and invert than to click the ninety-six you do.
These three operations sound interchangeable and produce completely different results. Choosing wrongly is the most common reason somebody runs a PDF tool twice.
| Tool | What you get | The question it answers |
|---|---|---|
| Extract Pages | One new document containing only the pages you picked. The rest are simply not in it. | "I need pages 4 to 9 to send to somebody." |
| Remove Pages | One document containing everything except the pages you picked. | "This document is fine apart from these three pages." |
| Split PDF | Several documents, between them normally holding every page of the original. | "This one file should have been five files." |
The rule of thumb: if you want one file out, you want Extract or Remove. If you want several files out, you want Split. And between Extract and Remove, pick whichever one lets you click fewer tiles - extracting 6 pages out of 200 and removing 194 of them produce the identical document, and one of them takes six clicks.
One genuine difference separates them. Extract in One PDF per page mode overlaps with Split in Every page mode, but only Extract lets you choose which pages get their own file. Splitting every page is all or nothing.
A page in a PDF is a small dictionary. It says how big the sheet is, how it is rotated, and which objects it needs in order to be drawn: a content stream of drawing instructions, the fonts those instructions name, any images they place, the color spaces those images are defined in, and any annotations sitting on top.
Extraction creates an empty document and copies the pages you picked into it, then follows every reference each page holds and copies the object on the other end as well. Nothing is rasterized. Nothing is decoded, nothing is re-encoded, and the image on your extracted page is the same compressed stream that sat in the original file, moved into a new container.
This is the number that catches people out. You extract three pages from a large document expecting a small file, and get something a third of the original size.
A page brings its dependencies with it, and those dependencies were shared. If the report embeds a full corporate font family - four weights, complete character sets, 9 MB - and your three pages use two of those weights, both of those font programs travel into your extract in full. They were 9 MB serving 200 pages. Now they are 5 MB serving three. Add a letterhead image placed on every page and a color profile, and a three page extract lands at 12 MB while being entirely correct.
The counter-case is a scan, where each page owns one image and shares nothing. Extract three pages from a 200 page scan and you get almost exactly three pages' worth of bytes, because there was never anything shared to bring along.
Some writers subset embedded fonts and some do not. A document produced by a careful tool carries only the glyphs it actually uses, so the extract stays small, while a document exported by an application that embeds full fonts hands your three page extract the entire typeface. You cannot tell which you have by looking at the pages. The file size afterwards is the only tell.
Extracting is a good privacy operation and a poor sanitization one, and the difference matters when the extract is going outside your organization.
The metadata control on the right handles the last one in a click. For the others, open the extract before sending it and press Ctrl+A in a viewer: anything that highlights is text somebody else can copy, including text you cannot see.
If you click page 9, then page 2, then page 5, the extract contains pages 2, 5 and 9 in that order, not in the order you clicked. This is deliberate and it is worth explaining because it occasionally surprises people who expected the clicks to be an ordering instruction.
Selection order is invisible. Once four tiles are marked, nothing on screen records which one you marked first, so an output built from click order would be built from information the user cannot see and cannot check. Two people making the same selection differently would get different documents. Document order is the one arrangement that is always visible, always reproducible, and always what a page range means.
It also keeps the typed field honest. Typing 9, 2, 5 means the same thing as typing 2, 5, 9, which is exactly how everybody reads a page range in every other context.
If you need the pages in a different order, extract first and then run Organize PDF on the result. That tool exists specifically for ordering, with drag and drop on the same kind of board, and doing it in two clear steps beats one tool that quietly does two jobs.
Extraction is what people reach for when a document is too sensitive to send whole. One page of a bank statement, for a landlord with no business seeing the rest of the account. The results section of a medical report. A single clause of a contract for a lawyer. The operation exists precisely because the rest of the file should not be seen by the person who is receiving the part you send.
Every other online extractor achieves that by receiving the whole file. Your entire bank statement is uploaded so that you can send one page of it, and it sits in a temporary directory until a cleanup job runs. The privacy motive that made you extract is defeated by the tool you used to do it.
This tool has no server in the path. The file is read by JavaScript on your own machine, the extract is built in your own memory, and it is handed back as a download. Load this page, disconnect from the internet, and it still works - which is the only demonstration that means anything. The single network call a run makes counts the operation and carries the tool name.
The document you are extracting from never enters a log, a queue or a backup. No policy promises it is deleted afterwards, because it is never transmitted in the first place and there is nothing anywhere to delete. Those are very different guarantees. Only one of them survives a breach at the other end.
The stated limits are 100 MB per batch on the free tiers and 500 MB on Premium, but the real ceiling is your device: the source document is held parsed in memory while the extract is built alongside it. A failed run is refunded automatically, so a file too large for your phone costs nothing to find out about.
| Symptom | Cause | Fix |
|---|---|---|
| The extract contains the wrong pages | The printed page numbers do not match the file's page numbers - normal for anything with a cover or contents section. | Pick by looking at the tiles. The number in a tile corner is always the true position in the file. |
| A three page extract is enormous | Embedded fonts and shared images were copied in full. | Expected. Explained in the engineering section above. Compress the extract separately if the size matters. |
| Pages came out in the wrong order | Extraction always uses document order, never click order. | Run Organize PDF on the extract and drag them into place. |
| Text is selectable in a scanned extract | The page carries an invisible OCR layer, which travels with it. | Expected behavior. Check with Ctrl+A in a viewer before sending anything sensitive. |
| Bookmarks are missing | The outline is dropped deliberately. | Nothing to fix. Most of its entries would point at pages that are not in the extract. |
| The browser blocked several downloads | One PDF per page mode produces many files. | They arrive as a single ZIP automatically above one file, so this only appears if downloads are heavily restricted. Allow downloads for this site. |
Drop the PDF into the zone at the top of this page. Every page appears as a tile. Click the pages you want, shift-click to take a whole run, or type them as 1, 4-9, 22, then choose whether you want one document or one file per page and press Extract pages. It is free, there is no watermark, and no account is needed for your first run.
No. The file is read by JavaScript on your own machine and the extract is built on your own processor. Load this page, disconnect from the internet entirely, and it still works, because there is no server-side step to depend on. This matters here more than on most tools: people extract pages precisely because the rest of the document is the part nobody else should see.
Click that single page on the board and press Extract pages. You get a one page PDF containing exactly that sheet, with its page size and rotation preserved and its content copied across without being re-encoded. The original on your disk is never modified.
Because a page brings its dependencies with it. Embedded fonts, a letterhead image placed on every page and color profiles were shared across the whole document, and every one your extracted pages use is copied across in full. A three page extract from a report with a 9 MB embedded font family can easily run to several megabytes. Scans differ. Each page owns its own image and shares nothing.
No. Pages are copied as compressed streams and are never decoded or re-encoded, which means the text, the vectors and the images in your extract are bit-identical to the original. Page dimensions and rotation are preserved exactly, including documents that mix page sizes on purpose.
Yes, in two ways. Click page 10 and shift-click page 25 to take everything between them, or type 10-25 into the Pages to extract field. The two stay synchronized in both directions, so you can type a range, add an extra page with a click, and read the corrected range straight back out of the field.
No, they come out in document order. Selection order is invisible on screen, so building the output from it would use information you cannot see or check, and two people making the same selection would get different files. If you need a specific order, extract first and then drag the pages into place with Organize PDF.
Extracting gives you one new document containing only the pages you picked, and the rest simply are not in it, which is the whole difference between the two operations. Splitting gives you several documents which between them normally hold every page of the original. One file out means extract. Several files out means split.
Not directly. Encryption covers the object streams themselves, so no page of the file can be read, let alone copied into an extract, until the password is gone. The card reports it as locked straight away. Remove the protection with Unlock PDF first, on a document you are authorized to open.
By default yes, which includes the title, author and creator fields your original application embedded, and those sometimes contain a username or an internal file path. Set the metadata control to Strip all and every one of those fields is written empty. Page level content is different. Invisible OCR text and annotations always travel with the page, whatever you do to the metadata.
Yes. Set Output to One PDF per page and each selected page becomes an independent document, named after its page number and delivered as a single ZIP so your browser is never asked to download thirty files in a row. That is the right choice when each sheet is a separate record, such as a set of certificates or individual policy schedules.
No. The extract contains the pages you picked and nothing else - no stamp, no logo, no trial banner across the first sheet, and no attribution line you are required to keep. Commercial use is free, including a page you pull out of a report and bill a client for.
Yes. The page tiles are sized for a fingertip rather than a mouse pointer, and on a small screen the page list field usually beats tapping: type 4-9 and the tiles select themselves. Watch the file size though. A phone tab has far less memory to work with than a desktop one.
Every page of the document if you want, and there is no limit on the number of separate ranges. If you are keeping most of the document there is a faster route: click only the few pages you do not want, then press Invert. Four clicks beat ninety-six. Remove Pages reaches the same result from the other direction.
All of these run the same way: your file is read by JavaScript on your own machine, the work happens on your processor, and the result is handed back to you as a download that never existed anywhere else.
Find the apps that slow your store down, by name and in milliseconds, then answer shoppers' doubts with trust badges under Add to cart. Both are free, with no card and no access to your store admin.
Paste any page and get its three Core Web Vitals, the apps costing time, like Klaviyo or Judge.me, and the Shopify or WooCommerce fix.
Show the promises that settle doubts, from free shipping to a money-back guarantee, right under the Add to cart button. One line in your theme.
Change the link after printing and count the scans. 5 tracked codes on a free account, static codes unlimited.
Anonymous visitors get one operation per 12 hours, per network address. That is not a sales tactic, it is the only lever we have: this tool runs on your processor, not ours, so the run itself earns us nothing. An account tells us who you are, and the ceiling goes up the moment you have one. Your allowance resets in 12 h.
No credit card, no trial period, no sales call. Join with Google, GitHub, Facebook or email.
Free accounts get two operations per 12 hours, and yours reset in 12 h. Waiting costs you nothing and is a completely reasonable choice. If PDF work is a weekly job rather than a one-off, the counter is the wrong tool for you and Premium takes it off the page.
Cancel anytime. Everything you have already produced stays yours either way.