You have a 30-page report and only need pages 5 and 12 to send to a client. Or just the first page of a contract for a quick reference. Or the last page of a signed agreement for your records.
This guide covers how to extract specific pages from a larger PDF, when extraction is the right approach (vs. splitting or deleting), and how to handle the common scattered-pages scenarios.
Extract vs Split vs Delete — what's different
Three operations all sound similar but produce different results:
| Operation | What it does | Output |
|---|---|---|
| Extract | Pulls out specific pages, leaves originals intact | New file with chosen pages only |
| Split | Breaks a PDF into multiple pieces | Multiple files based on the rule |
| Delete | Removes specific pages from the original | Smaller version of the original |
If you want just page 5 as a separate PDF, extract. If you want two PDFs — one with pages 1-4 and one with 5-30, split. If you want to remove page 5 and keep 1-4 and 6-30 in one file, delete.
For "I need page 5", extract is fastest and cleanest.
Common extraction scenarios
Scenario 1: Just the first page
A surprisingly common request — first page of a passport, a cover sheet, a title page:
- Open a page-extraction tool (e.g., PDFGrover's Extract Pages).
- Upload your PDF.
- Specify pages:
1. - Extract.
Output is a single-page PDF containing only your first page.
Scenario 2: First N pages
For previews, summaries, or sending the executive summary section:
- Specify pages:
1-5(the first 5 pages). - Extract.
Output is a 5-page PDF.
Scenario 3: Last page
Often needed for signed contracts (the signature page is usually last):
- Open the original PDF and note the total page count (shown in any reader).
- Specify the last page number directly: e.g., for a 30-page PDF,
30. - Extract.
Some tools support a last keyword or a "last page" button so you don't need to count manually.
Scenario 4: Specific scattered pages
You need pages 3, 7, 12, and 25 from a 30-page document:
- Specify pages with commas:
3, 7, 12, 25. - Extract.
Output is a single PDF containing those four pages in the order specified.
Scenario 5: A range plus extras
Pages 1-3 (introduction) plus pages 25-26 (conclusion):
- Specify:
1-3, 25-26. - Extract.
Output is a 5-page PDF (pages 1, 2, 3, 25, 26).
Scenario 6: Every other page
Useful for double-sided scans where you only need the front (odd-numbered) pages:
- Most tools don't have a built-in "every other page" feature.
- Manually specify:
1, 3, 5, 7, 9, 11...(tedious for long documents). - Alternative: use Split PDF → Every page → then merge selected outputs.
For long documents needing every other page, scripted approaches (Python, command-line tools) are faster.
Step-by-step in detail
For a typical extraction workflow:
Open your source PDF to confirm the page numbers you want. Page numbers in the file may differ from page numbers shown on the page itself (e.g., the cover counts as page 1 even if it isn't numbered).
Open an extraction tool — browser-based tools like PDFGrover's Extract Pages work without uploading the file (for files under the size threshold) and don't require signup.
Upload or drag the PDF into the tool.
Enter the page range in the format the tool accepts:
- Single page:
5 - Continuous range:
5-10 - Discrete pages:
5, 7, 9 - Combination:
5-10, 12, 15-17
- Single page:
Confirm the order matters. Most tools extract in the order you specify, so
5, 3, 7produces an output with page 5 first, then 3, then 7. If you want them in source order, list them sorted:3, 5, 7.Click Extract and download the result.
Verify by opening the extracted file. Confirm you have the right pages in the right order.
When the page numbers in your file don't match what you see
Two different page numbers can appear:
File page number — the position in the PDF (page 1 = first page of the file, regardless of what's shown on it).
Visible page number — what's printed on the page itself (e.g., "page 7" of a chapter).
Extraction tools work with file page numbers. So:
- A document with no cover where "page 1" shows on the first sheet → file page 1 = visible page 1. ✅ Easy.
- A document with a cover (no number) + TOC (Roman numerals i, ii, iii) + body (Arabic 1, 2, 3) → file page 5 = visible body "page 1". Confusing.
To check: open the PDF, scroll to the page you want, look at the "Page X of Y" indicator in your reader's status bar — that's the file page number.
If you specify page numbers that match the visible numbers when the file uses front-matter, you'll extract the wrong pages. Always use file page numbers.
When to extract vs other approaches
Extract when:
- You want a small subset of pages
- You need to send specific pages without the rest
- You're picking scattered pages from a long document
Split instead when:
- You need many smaller files (one per page, or one per range)
- You're separating sections (e.g., chapters)
- The rule is uniform (every page, every N pages)
Delete instead when:
- You want everything except a few pages
- The "keep" set is much larger than the "remove" set
- You want to keep the original layout structure
A practical example: a 50-page report where you want to send everything except 3 internal pages. Use delete pages 23, 31, 47 rather than extract pages 1-22, 24-30, 32-46, 48-50.
Combining extractions from multiple PDFs
If you need pages from several PDFs combined into one:
- Extract from each source PDF separately, producing several small PDFs.
- Merge the extracted pieces into one final document.
Example workflow:
- Extract page 5 from ContractA.pdf → contractA-page5.pdf
- Extract pages 1-2 from ContractB.pdf → contractB-pages1-2.pdf
- Extract page 12 from ContractC.pdf → contractC-page12.pdf
- Merge all three into one combined PDF
For doing this often, scripted tools (Python's PyPDF2, command-line qpdf) are faster than the manual workflow.
File size considerations
Extracted PDFs sometimes look "too large" for the page count. Reasons:
- Embedded fonts inherit. All fonts used anywhere in the source are embedded in the extract, even if the extracted page only uses a few of them. Some tools subset; some don't.
- Compressed images stay compressed. A page with a 5 MB embedded photo extracts as a 5 MB PDF.
- PDF overhead. Even a single-page PDF has structural overhead — usually 10-50 KB minimum.
If your extracted PDF is unexpectedly large (e.g., 5 MB for a single page that should be tiny), run it through a compressor to shrink.
Common mistakes
Confusing file page numbers with visible page numbers. As covered above. Always use file page numbers.
Specifying ranges in the wrong format. Tools differ — some accept 5-10, some 5..10, some 5,6,7,8,9,10. Check the tool's help if your range gets rejected.
Forgetting to deduplicate. If you list 5, 5, 7, some tools include page 5 twice in the output. Deduplicate your list before extracting.
Extracting then editing the original. The extracted file is independent. If you later edit the source, the extract still shows the OLD content. Re-extract if you need updated pages.
Sending the original by mistake. Save the extracted file with a clear name (contract-signature-page-only.pdf), not the same name as the source. Recipient confusion happens when both files have similar names.
Trying to extract from a password-protected PDF. First unlock, then extract.
Quick reference
| Need | Specify |
|---|---|
| First page only | 1 |
| Last page only | The page number (count the pages first) |
| First 5 pages | 1-5 |
| Last 3 pages of a 30-page PDF | 28-30 |
| Pages 5 and 7 only | 5, 7 |
| Pages 1-3 + page 12 + pages 25-30 | 1-3, 12, 25-30 |
| Even-numbered pages | 2, 4, 6, 8, 10... (manual) |
Summary
Extract is the right operation for "I want some pages from this PDF." It produces a new file containing only the chosen pages, leaving the original untouched.
Quick decision: extract for picking a few pages; split for breaking up by rule; delete for removing a few while keeping most.
PDFGrover's Extract Pages tool runs in your browser for files under 100 MB — no upload, no signup, no watermark. Specify pages, click extract, download. Combine with Merge PDF when you need pages from multiple sources in a single output.