Why does every page-tool URL show the same form?

Because it is one pass over the document. Reordering, deleting, rotating, numbering, watermarking and writing metadata are all decisions taken while the pages are copied into a new file, and doing them in six separate runs would rewrite the same document six times. Each job has its own address because each is a different thing to search for, and each page names its job in the heading — but the controls beneath are, honestly, the same controls.

How do I delete pages here?

Leave them out of the order box. There is no separate delete control: on a ten-page document, typing 2-9 keeps eight pages and drops the first and the last. Reordering and removing are the same pass, so you can do both at once.

Can I repeat a page, or reverse the document?

Neither. Repeated numbers are dropped, so 1, 1, 2 produces a two-page file; extract the page separately and merge it back in if you need it twice. And a backwards range such as 10-1 is refused with 1-10 suggested, because a range that silently reversed itself would be a different operation from the one you typed. To reverse a document, write the numbers out in the order you want.

Is the result checked before I download it?

Yes, twice. The saved file is loaded again and its page count compared with the number of pages you asked for, and every page’s rotation is checked to be a whole quarter turn. A mismatch fails the run instead of producing a file.

One unencrypted PDF of up to 150 MB, and at most 2,000 pages in any single range you type. An encrypted file is refused with advice to remove its password locally first. A watermark is capped at 80 characters and accepts basic Latin only.

Does it change the quality of my pages?

No. Pages are copied whole rather than redrawn, so nothing is rasterised: text stays selectable, vectors stay sharp, images are the same images, and a page of a different size in the middle of the document keeps its size.

PDFPage tools

Organize PDF pages

Reorder, remove, rotate, number, watermark, and label a PDF in one local pass.

On-device prototype

Source PDF

Signature and page count checked locally

About this PDF page organiser

Choose a PDF, type the page numbers in the order you want them — 3, 1-2 — set a rotation, a watermark, page numbers and document metadata as needed, and apply. Everything happens in a single pass: the listed pages are copied into a new document in your order, anything you leave out is dropped, and the saved file is reopened and checked before a download is offered.

This is one pass over one document, which is why all six jobs share a screen: reordering and deleting are the same box, and rotating, numbering, watermarking and writing metadata all happen while the pages are being copied. Doing them separately would mean four rewrites of the same file. It takes one unencrypted PDF of up to 150 MB, copies pages whole so nothing is rasterised and no quality is lost, and verifies the result twice — the page count against what you asked for, and every page’s rotation against a whole quarter turn.

Runs in this tab — no uploadFree, no account, no watermarkKept for use with the network off

Step by step

1Choose the PDFOne file of up to 150 MB, checked on its first five bytes. Its page count is read straight away and the order box is filled in with the whole document, so you edit down from a correct starting point.
2Say which pages, in which orderCommas separate entries and a hyphen makes a range. The order you type is the order you get, and any page you leave out is left out of the file — which is how deleting works here.
3Add anything else that belongs in the same passA rotation applied to every page, a watermark of up to 80 characters, centred page numbers, and a title, author, subject and keywords for the document.
4Apply and downloadThe work runs in a Web Worker. The saved bytes are reopened, the page count compared with what you asked for and every rotation checked to be a multiple of 90, before the file is offered as edited-pages.pdf.

Why one screen rather than six

Every job on this page is a decision made while pages are being copied from one document into another. Reordering is the order of the copy; deleting is a page not copied; rotating sets a property on the copy; numbering and watermarking draw onto it; metadata is written on the document that receives it. Splitting them into six tools would mean rewriting the same file up to six times, and each rewrite is another chance to lose something.

Each of the six also has an address of its own, because "rotate PDF" and "add a watermark to a PDF" are different things to search for and a single page with one title cannot answer both. What those addresses change is the heading and the explanation; the form is the same form, and it says so.

The page order box, in detail

Commas separate entries, a hyphen makes a range, and spaces around a hyphen are fine. Any number outside your document is refused with the valid range quoted back, and a single range may cover at most 2,000 pages.

Two refusals are deliberate. A page listed twice appears once — repeats are dropped — so if you need a page twice, extract it separately and merge it back in at the position you want. And a backwards range such as 10-1 is refused with the right way round suggested, rather than silently reversing itself: a range that quietly means the opposite of what it says is a worse outcome than an error message.

What the pass preserves, and what it does not

Pages are copied, not redrawn. Page content, page size and orientation are untouched, text stays selectable, and vector drawings stay sharp at any zoom.

What does not follow is the material around the pages. The output is a newly created document, so document-level metadata from the source does not come across: what gets written is whatever you type into the title, author, subject and keywords boxes, plus a producer of "Browser Tools". Bookmarks, form behaviour, attachments and certification state are not guaranteed either — the scope is visible page content and order, and the page states that at the bottom rather than in a footnote.

Where it runs

Every response from this site is served with a Content Security Policy whose connect-src directive is set to none. That is not a promise in marketing copy; it is an instruction to your browser, and the browser is the one enforcing it. While this page is open it cannot open a connection to anywhere — not to this site, not to anyone else — so there is no path by which your document could be uploaded, not by a bug, not by a future change, not by a script that should not be here. e2e/egress-proof.spec.ts asserts the served header and then tries to send data out by every route a page has and requires each attempt to be refused, and it runs on every build.

This is one of the pages held in the service worker cache, so after a first visit its code is already on your device.

The second half of the guarantee is that the code has nowhere to send anything from. lib/tools/local-source-policy.test.ts reads every source file under lib/tools, workers, components and app on every test run and fails the build if any of them contains fetch, XMLHttpRequest, WebSocket, EventSource, sendBeacon or a peer connection, or even a remote address written down in a comment. A tool here cannot acquire an upload by accident, because the file that would have to contain one cannot be committed.

Questions people ask

Why does every page-tool URL show the same form?: Because it is one pass over the document. Reordering, deleting, rotating, numbering, watermarking and writing metadata are all decisions taken while the pages are copied into a new file, and doing them in six separate runs would rewrite the same document six times. Each job has its own address because each is a different thing to search for, and each page names its job in the heading — but the controls beneath are, honestly, the same controls.
How do I delete pages here?: Leave them out of the order box. There is no separate delete control: on a ten-page document, typing 2-9 keeps eight pages and drops the first and the last. Reordering and removing are the same pass, so you can do both at once.
Can I repeat a page, or reverse the document?: Neither. Repeated numbers are dropped, so 1, 1, 2 produces a two-page file; extract the page separately and merge it back in if you need it twice. And a backwards range such as 10-1 is refused with 1-10 suggested, because a range that silently reversed itself would be a different operation from the one you typed. To reverse a document, write the numbers out in the order you want.
Is the result checked before I download it?: Yes, twice. The saved file is loaded again and its page count compared with the number of pages you asked for, and every page’s rotation is checked to be a whole quarter turn. A mismatch fails the run instead of producing a file.
What are the limits?: One unencrypted PDF of up to 150 MB, and at most 2,000 pages in any single range you type. An encrypted file is refused with advice to remove its password locally first. A watermark is capped at 80 characters and accepts basic Latin only.
Does it change the quality of my pages?: No. Pages are copied whole rather than redrawn, so nothing is rasterised: text stays selectable, vectors stay sharp, images are the same images, and a page of a different size in the middle of the document keeps its size.