Extract Pages

Extract Pages From a PDF

You have a document. You need part of it. Type which pages, get a new PDF containing exactly those, named after the original with -extracted.pdf on the end. The source file on your disk is untouched.

Order follows the document, not your typing

Pages come out in the order they appear in the source. Typing 9,2,5 gives you pages 2, 5 and 9 in that sequence — the selection says *which*, not *what order*.

If you need a different order, extract first and then retype the sequence in Organize PDF. Keeping the two operations separate is deliberate: it means a selection can never accidentally duplicate or drop a page.

Selections for common jobs

7            one page
12-19        a chapter
1,3,5,7      alternating pages from a duplex scan
1-2,45-48    cover plus the appendix
last         whichever page ends the file
odd          all odd-numbered pages
even         all even-numbered pages

The odd and even keywords exist mainly for one job: a double-sided document fed through a single-sided scanner produces all the fronts, then all the backs. Extracting odd and even separately is the first step in putting it back together.

Numbers refer to positions in the file, counting from 1. A report whose printed page 1 is the fifth sheet — after a cover, a blank, a title page and a contents page — needs 5 for that page, not 1.

Three tools, one decision

Extraction, splitting and deletion all produce a smaller PDF. Which one you reach for depends on how you are thinking about the problem:

The practical test is which list is shorter. Keeping 4 pages out of 300? Extract. Dropping 4 pages out of 300? Delete. Doing it the other way round means typing a range that spans the whole document.

What comes with the pages, and what breaks

The page content is copied exactly — same text, same images, same compression, no quality loss. Annotations and link rectangles attached to the extracted pages come across too.

Two things behave differently:

Links to pages you left behind. A cross-reference pointing to page 200 has nothing to point at in a four-page extract. Readers treat it as a dead link rather than showing an error.

File size. An extract is rarely proportional to its page count. Shared resources — embedded fonts, a logo used on every page, an ICC colour profile — get copied along with the pages that reference them. Pulling one page out of a heavily designed brochure can produce a file most of the size of the original.

A note on what this is for

Extraction is the polite way to share part of a document. Sending someone a 300-page contract when they asked about clause 14 is a small act of hostility; sending them pages 41 to 43 is not.

It is also the right first step before almost anything else here. Rotating, watermarking or compressing three pages is faster than doing it to three hundred and then throwing most away.

Encrypted PDFs are refused at the upload stage — decrypt with Unlock PDF first. Files up to 100 MB are accepted and nothing leaves your browser.