Extract PDF pages — keep exactly the pages you need, in the order you want
Extracting pages from a PDF sounds simple, and on the surface it is — pick the pages, get a smaller PDF. The friction is in what "pick the pages" actually means: a range syntax that supports comma-separated single pages and dash-separated ranges, with order preserved exactly as you typed it. Pair that with the fact that extract and split look identical at first glance and you have one of the most-confused operations in any PDF toolbox. Here is what extract does, when it is the right tool, and how to spec the pages you actually want.
What extract pages does that split doesn't
Extract and Split look related but produce different shapes of output. Split takes one PDF and produces several — one output file per range you specify. Extract takes one PDF and produces one new file — a combined output with only the pages you wanted, glued together. If you want pages 1-3 plus pages 8-10 as a single 6-page deliverable, you want Extract. If you want pages 1-3 and pages 8-10 as two separate files, you want Split.
The other practical difference: Extract honors the order you list. Type "5, 1, 3" and you get pages 5, 1, 3 in that exact order — useful for reordering small selections without going through the visual sort tool. Split, by contrast, always emits files in document order regardless of how you typed the ranges.
How to spec the pages you want
The range syntax is short enough to learn in a minute, expressive enough to cover every reasonable selection. Each comma-separated chunk is either a single page or a dash-separated range. Order is preserved as typed.
- 1-5 Pages 1, 2, 3, 4, 5 in order. A simple range.
- 1, 3, 5 Pages 1, 3, 5 — every other page from the top, by hand.
- 1-3, 8-10 Pages 1, 2, 3, 8, 9, 10 — two ranges combined.
- 5, 1, 3 Pages 5, 1, 3 — out-of-order on purpose. Output has those pages in that exact order.
- 1-3, 7 Mix range with single — 1, 2, 3, 7.
- 1-N Whole document. (N is the last page; the form accepts the literal letter N as shorthand.)
Common real-world uses
Extract earns its keep in scenarios where you need to slice content but produce a single coherent deliverable:
- Send just the relevant pages of a contract. Extract the signature page + amendment clauses; share the 4-page extract instead of the 80-page master.
- Build a study deck from a textbook. Extract pages 12-18 + 47-52 + 110-115 into one 18-page PDF you can annotate without dragging around the whole book.
- Re-order a deck for a different audience. Pull slides 3, 8, 1, 6, 12 in that order to build a 5-slide talk track from a 30-slide master deck.
- Isolate the page you actually need. Extract just page 47 of a 500-page report — open Extract, type "47", download. Faster than scrolling and screenshotting.
- Build a cover letter + relevant resume sections. Extract your cover letter page, plus the resume's relevant experience block, into one 2-page submission.
Things that will catch you out
The friction points new users hit most often:
- Page numbers in the PDF vs the document's printed page numbers. Extract uses physical page positions, not printed page numbers. If your document has a 5-page front matter (Roman i-v) before the body (1, 2, 3…), "extract pages 1-3" gives you the front matter, not the chapter you wanted. Add the front-matter offset (i.e. extract "6-8" for the first three numbered pages).
- Trailing commas and spaces. Most PDF tools choke on trailing whitespace or commas. Ours doesn't, but it is worth typing the range cleanly so the next tool you use doesn't break.
- Overlapping ranges. If you type "1-5, 3-7" you get pages 1, 2, 3, 4, 5, 3, 4, 5, 6, 7. The same pages appear twice. Usually a typo — review the spec before clicking apply.
- Out-of-range page numbers. Asking for page 200 in a 50-page document produces an error. We show the document's page count near the input so you can sanity-check the spec.
Why extract preserves quality
Like rotate, extract is a structural rather than rendering operation. We parse the input's page tree, copy the page objects you selected to a new PDF, and rebuild the cross-reference table. Page contents are transplanted byte-for-byte: text stays selectable, vector graphics stay sharp, embedded fonts come with their pages, scanned images keep their resolution. The output file is the natural size for the page count you extracted — typically a fraction of the input, scaled by the fraction of pages kept.
Annotations, hyperlinks, and form fields on extracted pages survive the extract. Hyperlinks that pointed to a different page of the original PDF are remapped: if the original page 5 linked to page 12, and you extracted only page 5, that link now points to nothing (the target is no longer in the file). Hyperlinks to external URLs survive unchanged. Bookmarks are pruned to match the new page set.
Limits and compatibility
On the free web tool, extract handles PDFs up to 100 MB with no page-count cap. Processing runs entirely in your browser via pdf-lib; nothing is uploaded. Output is byte-compatible with every PDF viewer since Acrobat 5. The extracted PDF is structurally a fresh, independent file — no link to the original, no embedded reference. You can rename it, share it, or pass it through further processing with no concerns about the source.
Common next steps: pair with Merge to combine extracts from multiple source PDFs into one bundle, or with Page Numbers to add fresh numbering to the extracted set (the original page numbers no longer correspond to the page positions in the new file).