Extract PDF pages

This is the tool for the situation where a whole file is more than the recipient should have: one page of a contract, the appendix without the correspondence, the relevant three pages of a fifty-page report. Everything you did not name simply is not in the file that comes out.

Worth being precise about what that means. The pages you left behind are genuinely absent from the new document — this is not hiding or cropping, and there is nothing to recover from the result. What it is not is redaction: anything on a page you did keep is still fully there, including text under a black rectangle.

Typing 1-3, 7, 10-12 into a box and pressing a button is easy to get subtly wrong, and the mistake is invisible until somebody opens the file. So the selection is summarised in plain form — pages 1–3, 7, 10–12 — before you commit, and a page number that does not exist is refused with the document’s real length rather than quietly dropped.

Extracting keeps the pages you name and produces one document. Splitting takes the whole file apart and produces many. If you want pages 4 to 6 of a forty-page report to send to somebody, this page does it in one step; splitting would give you forty files to sort through.

The rule of thumb: extract when you know which pages matter and the rest does not. Split when every page needs to become its own document, which is the receipts-and-statements case rather than the send-me-the-appendix case.

Each page arrives at its original size and orientation, with its text still selectable and its vector artwork still sharp — the pages are copied, not redrawn, so there is no quality setting and no quality loss. A page that was A3 landscape in the source is A3 landscape in the result, even if every other page was A4.

What does not come is anything that pointed at the document as a whole. Bookmarks, the outline and form fields refer to pages by position, and those positions no longer exist; links between pages you kept survive, links to pages you left behind do not. Page numbers printed *on* the page are part of the page, so an extracted page 7 still says 7 — which is usually what you want and occasionally confusing.

To take three pages out, the document has to be opened and understood in full — including the parts you are removing. That happens in your browser, so nothing about it leaves your device, but it is worth stating because the point of this tool is usually that the rest of the file is confidential.

The consequence to be precise about: the result contains only what you named, and the pages you dropped are genuinely gone from it. There is nothing hidden in the output to recover. That is a claim about the file you send, not about a promise that the rest was never processed.

Extraction works on whole pages, which is fine when the document is already organised the way you need. It is the wrong tool when the sensitive part and the part you want to send are on the same page — one paragraph of a contract, a figure with a client name in its caption.

There is no version of this that solves that, here or anywhere: removing something from within a page means editing the page, and a rectangle drawn over text leaves the text underneath. If part of a page must not travel, the honest routes are a proper redaction tool that removes the content, or retyping what you need.