How to Split a PDF and Extract Pages

When Splitting Helps

• Sending the relevant part, such as the two pages of a 48-page contract that need a signature.
• Upload forms that limit file size or page count.
• Separating a batch scan: 30 receipts fed through a scanner in one go end up in one file. If each receipt is one page, Extract All Pages gives one file per receipt.
• Preparing pages for another tool: the PDF to JPG tool converts every page into one ZIP or saves pages one click at a time, so extracting pages 5–12 of an 80-page report first gives a ZIP of 8 images instead of 80 (see PDF pages to JPG or PNG).

To fix page order or delete a few pages, see How to Reorder, Rotate and Delete PDF Pages instead.

Two Modes, Three Kinds of Result

What you download depends on the tab and on how many pages are selected. For report.pdf:

So 1-3, 5, 8-10 gives one 7-page file, not three. For separate sections, extract them one after another: the document stays loaded after a download, so enter 1-3, download, enter 5, download again. Every multi-page extraction is named report_extracted.pdf, and browsers usually add a number to repeated names, so rename the files as you go.

Step by Step

1. Open Split PDF and choose one PDF in Choose PDF File or Drag & Drop Here. If you drop several, only the first is used.
2. Wait while thumbnails of all pages are drawn. The file name, size and page count then appear above the tabs, and all pages start out selected.
3. On Custom Range Extraction, type into Page Range (e.g. 1-3, 5, 8-10) or click thumbnails. The box starts with the full range, such as 1-40, so clear it or click Deselect All first. Select All restores every page, and a counter shows, for example, "7 of 40 pages selected".
4. Or switch to Extract All Pages, which ignores the selection.
5. Click Split PDF & Download; Change File starts over.

Clicking thumbnails rewrites the range box compactly: pages 2, 3, 4 and 9 become 2-4, 9. Thumbnails are drawn at low resolution, so check the content in the downloaded file.

Writing Page Ranges

Numbers are positions in the file, starting at 1, as on the badges under the thumbnails.

• Separate items with commas; use a hyphen for a range. Spaces around them do not matter.
• Order and repeats do not matter: 8-10, 1-3, 2 gives pages 1, 2, 3, 8, 9, 10 in their original order.
• 10-8 is read as 8 to 10, and 38-45 in a 40-page file stops at 40. A single number beyond the last page, or 0, is ignored.
• Open ranges such as 5- or -3 are ignored; write 5-40.
• Only the plain hyphen counts. An en dash pasted from a word processor, as in 1–3, selects only page 1, and so do 1 3 and 1;3.

Printed numbers often differ from positions. If a report has a cover and four front-matter pages (i–iv) before page 1, printed page 12 is file page 17 (1 + 4 + 12). Check the highlighted thumbnails before downloading.

Why the Pieces Can Add Up to More Than the Original

Fonts and images used on many pages, such as an embedded font or a header logo, are usually stored once in the original and shared by every page. Each new file needs its own copy of every resource its pages refer to. pdf-lib copies these resources as they are: it does not shrink fonts to the characters on the copied page or drop entries a page lists but does not use. A single page can therefore carry a font subset made for the whole document, and in files where all pages share one resource list, every font and image in that list.

In our test, we made a 40-page A4 text report with Python's ReportLab (two DejaVu fonts embedded as subsets, a 900 × 240-pixel JPEG logo on every page) and split it in Node.js with pdf-lib 1.17.1, the library version the LeedyBox tool loads, using the same calls:

A scan-like file behaved differently: ten A4 pages made with Pillow from 1,240 × 1,754-pixel greyscale JPEGs (150 dpi) came to about 2.6 MB, and the ten single pages together also to about 2.6 MB, because nothing is shared. Your numbers will differ, but in short:

• Fewer pages shrink scans roughly in proportion (our 7-page extract of the scan was about 1.8 MB); text documents keep a base size for their fonts and shared images.
• Several pages in one file usually take less space than the same pages as separate files.
• The tool stores the PDFs in the ZIP without compressing them, so the ZIP is about as large as the files it contains.

If a file is still too large, see How to Compress a PDF.

What the New Files Keep and Lose

Pages are copied, not re-rendered: text stays selectable, images are not recompressed, and page size and rotation are kept. Web links on the copied pages are kept too. What belongs to the document as a whole is not copied, and some page features depend on it:

• Bookmarks and page labels (custom numbering such as i, ii, iii).
• Title, author and other document properties. The new files list pdf-lib as the producer and the time of splitting as the creation date.
• Accessibility tags, the structure screen readers use for headings and reading order: even if the original is tagged, the new files are not.
• Internal links, such as table of contents entries, may lead nowhere if their target page was left out or if they use a named destination, which is stored at document level. Form fields may stop being fillable, because the form they belong to is also defined at document level.
• A digital signature does not stay valid: it covered the exact bytes of the original file, so a signature that still shows on a copied page proves nothing about the new file. Keep the signed original.

Splitting Is Not a Safe Way to Remove Sensitive Pages

In a test with pdf-lib 1.17.1, we extracted page 1 of a six-page PDF in which page 1 linked to page 5. The new file showed one page and a text extractor found only page 1's text, yet page 5's content was stored inside the file as unused data, readable by anyone who inspects the file's internals. Tables of contents and cross-references can pull in pages this way. Content hidden on a copied page, such as text under a black box or outside the visible area, is copied as well. For confidential material, use a PDF editor's redaction or sanitising function, or send page images made with PDF to JPG. An image holds only the drawn appearance of the page, not its text layer or hidden objects, though the text can then no longer be selected or searched.

Common Problems and Fixes

• "… is a PDF file that cannot be read by the browser or is corrupted." The file may be damaged or need a password to open; the tool has no password field. If permitted, save an unprotected copy in an application where you can enter the password.
• Thumbnails appear, but splitting fails with a message about encryption. The PDF opens without a password but is encrypted, which is how permission settings such as an editing or printing restriction are stored. The previews are drawn with PDF.js, but the splitting is done by pdf-lib, which, as the tool calls it, refuses encrypted files. The tool does not remove or get around this protection.
• "Please select at least one page to extract." Nothing valid is selected, for example after Deselect All or with an open range.
• Wrong pages. Look for an en dash, a missing comma or printed page numbers.
• Slow with long files. All thumbnails are drawn before the controls appear, and the progress bar stays at about a fifth until then, so a long file can look stuck. Extract All Pages also builds every file and the ZIP in memory. The tool sets no size cap of its own, but a phone may fail on a very large file where a computer copes.

Before You Send the Files

Check each file's page count, first and last pages and size, and give it a descriptive name such as contract_signature_pages.pdf. Keep the original. To join sections again, use Merge PDF. If the original has no printed page numbers and an extracted section should start at, say, 31, see Adding Page Numbers to a PDF. The page loads PDF.js, pdf-lib and JSZip from leedybox.com itself, but your file is read and split in your browser and is not sent to a LeedyBox server.

Frequently Asked Questions