How to Extract Specific Pages From a PDF Into a New File
Someone needs the payment terms. They are on pages 14 to 16 of a forty-page contract, and sending the whole contract means sending thirty-seven pages they did not ask for and should perhaps not have.
Extracting pulls out exactly the pages you name and saves them as a new document, leaving the original exactly as it was. Extract PDF Pages does that in your browser, so the contract itself never goes anywhere.
Extract or delete? The arithmetic decides
Extracting and deleting produce identical results by opposite routes. One names what stays; the other names what goes. Both end with a single shorter file.
The only question worth asking is which list is shorter to type.
| Situation | Faster operation |
|---|---|
| Keep 3 pages of 40 | Extract — type 14-16 |
| Remove 2 pages of 40 | Delete — type 7,19 |
| Keep roughly half | Either; pick whichever list is tidier |
Extracting three pages from forty means typing one range. Deleting the other thirty-seven means typing 1-13,17-40 — longer, and two more opportunities to mistype a boundary and quietly ship the wrong pages. The reverse is true when you are only dropping a couple of pages, which is what how to remove pages from a PDF covers.
If you want several files rather than one, neither is your operation — that is splitting.
Extracting, step by step
- Open Extract PDF Pages and choose your file. It stays on your device.
- Enter the pages you want to keep — a single page (
9), a list (2,7,14), ranges (14-16), or a mix (1,14-16,32). - Check the preview before exporting.
- Export. You get a new document containing exactly those pages, in the order you listed them.
Reversed ranges and page numbers past the end of the document are rejected with a message rather than silently producing something odd.
The order you type is the order you get
This is the one behaviour that surprises people, and it is occasionally very useful.
Type 5,2,9 and you get those three pages in that sequence — page 5 first. Extraction does not sort your list back into document order. Most of the time you will type ranges in ascending order without thinking about it and never notice. When you need a specific arrangement, you can build it here in one step rather than extracting and then reordering.
If you did not intend it, this is also the most likely reason an extracted file comes out in a strange order. Read your list back before exporting.
Page numbers mean sheets, not printed labels
14 is the fourteenth sheet from the front, not the page with “14” printed on it. In any document with an unnumbered cover and contents page, those differ by two — and a contract’s payment terms are exactly the kind of thing you do not want to be off by two about.
Count from the front, or click thumbnails and let the tool build the list.
Check the boundaries, not the middle
The characteristic extraction error is being one page out at an edge: a clause that begins at the foot of page 13, or a table continuing onto page 17.
So check the first and last pages of your result, not the middle. Does the first page start where the topic starts, or halfway through a sentence? Does the last page finish the thought? A section almost never begins exactly at a page boundary, which is why extracting 14-16 so often should have been 13-17.
When in doubt, take one page more on each side. A little extra context costs nothing and reads as thorough. A clause cut in half reads as careless, and invites a follow-up email asking for the rest.
Extracting is not redacting
Worth stating plainly, because extraction is often used precisely to keep information back.
Extracting genuinely removes the pages you did not select — they are not in the new document in any form. That part is sound.
What it does not do is remove anything sensitive on the pages you did keep. If page 15 contains both the payment terms and a name that should not travel, extracting page 15 sends that name. Two habits that look like they solve this and do not:
- Cropping hides, it does not delete. The content outside a crop box is still in the file and returns when the crop is reset — how to crop PDF margins.
- A black box drawn over text is a black box. The text underneath stays selectable and copyable.
If a kept page contains something that must not be seen, no operation on this site solves that. The only reliable answer is to not include that page.
What comes across, and what does not
The new file is rebuilt from the pages you chose, so:
- Page content, size, and orientation come through faithfully.
- Bookmarks are dropped. The rebuild has no outline pane to carry over.
- Digital signatures are invalidated — an extract is a different document, and a signature covers the document it signed. See electronic signature vs digital signature.
- Printed page numbers stay as they were. Your extract of pages 14 to 16 still says 14, 15, 16 — and if it should read 1, 2, 3, use Add Page Numbers to PDF.
- Internal links break where their target was not extracted with them.
- Encrypted files must be unlocked first — how to unlock a PDF when you know the password.
That page-number point catches people out on formal submissions, where an extract carrying the parent document’s numbering looks like a document with pages missing. Renumber before sending.
Naming the file you send
The extract arrives in someone’s inbox with no context around it, and its filename is the only thing telling them what it is. This is a thirty-second step that saves a round of email.
A name like contract-extract.pdf says nothing useful. acme-contract-payment-terms-p14-16.pdf tells the recipient which document it came from, which part they are looking at, and where it sits in the original — so if they need the surrounding context they can ask for it precisely.
Including the page range matters more than it looks. An extract is a document with a gap in its history: it says “page 14” at the foot of its first sheet while being the first sheet. Putting the range in the filename resolves that before anyone has to wonder about it.
If the extract is going into a formal bundle rather than an inbox, the naming convention is usually specified for you, and following it exactly is part of the job — see preparing a PDF bundle for submission.
Common uses
- Sending only what was asked for. The most common case, and the best reason: fewer pages is not just politer, it is less disclosure.
- Pulling one appendix out of a long report to circulate on its own.
- Building a new document from several sources — extract the relevant pages from each, then merge them.
- Isolating scanned pages in a mixed document so recognition runs only where it is needed.
- Getting a file under an email limit when the extra pages are the heavy ones. On a scan, where every page weighs about the same, dropping pages helps proportionally — unlike cropping, which saves nothing at all. Why is my PDF so large explains the difference.
Extracting from a document you did not make
Most extraction happens on documents someone else produced, which introduces two wrinkles worth knowing about.
The page count may not match what the document claims. A report whose contents page lists sections by printed page number is describing its own numbering, not the sheet order. Cross-reference once at the start — find the sheet carrying printed page 1 and note the offset — and every range you type afterwards is right.
A scan gives you nothing to search. Locating the section you need means scrolling rather than searching, because there is no text to match. If the document is long enough that this hurts, running OCR first turns twenty minutes of scrolling into one search; why a PDF is not searchable explains how to tell which kind of document you are holding.
Neither changes how extraction works. Both change how long it takes to work out what to type, which on a long document is the majority of the job.
Where this sits in the format
Pages hang off a structure called the page tree, specified in ISO 32000-2. Extraction builds a fresh tree containing only the pages you named and copies them into it, which is why the result is a proper standalone document rather than a view onto the original. The PDF Association publishes readable material on that structure.
Keep the original
The extract is a new file and your source is untouched, so a mistake costs a minute. Check the first and last pages, confirm the numbering makes sense to someone who has not seen the parent document, and send.