Free tool · PDF · Runs in your browser
Split a PDF
Short answer
Choose a PDF and split it into one file per page, per range (for example 1-3; 4-10), or every N pages, in your browser with no upload.
Nothing is uploaded. The PDF is read and written in your browser. Password-protected PDFs are not supported.
Key takeaways
- Pages and text are intact: all 7 test splits gave the right page count, the same page sizes and the same page text, checked by pdf.js.
- Each part holds only its own pages: in all 132 parts we counted the page objects in the file and the number equaled the pages shown.
- Text documents repeat their fonts in every part: Public Law 118-5 285 KB → 1.87 MB as 41 single pages (6.6x); in two halves it was 232 KB (81% of the original).
- Scans split almost without extra weight: Senate Document 105-10 3.06 MB → 3.02 MB as 67 pages; the 230 page NASA scan in 5 parts totaled 12.4 MB from 12.5 MB.
- Speed: 18 to 522 ms per split.
Measured: seven splits
Every part had the right pages, sizes and text in our 7 test splits, and each part holds only its own pages: the page count in the file equals the pages shown. The parts together can be much bigger than the original, because each part carries its own copy of the fonts its pages use: a 285 KB text law split into 41 single pages became 1.87 MB (6.6 times), while scanned documents came out about the same size (a 3.06 MB, 67 page scan became 3.02 MB).
Measured with the tool's own code in Node on six public domain US government PDFs: IRS Forms 1040 and W-9, Public Law 118-5, Senate Document 105-10, House Document 110-50, and a 230 page scanned NASA technical report (12.5 MB). Each result was opened with pdf.js, which compared the page count, every page size and the text of every page with the source pages.
| Document | Split | Parts | Original | Parts together | Together vs original | Smallest part | Largest part | Time |
|---|---|---|---|---|---|---|---|---|
| Public Law 118-5 | every page | 41 | 285 KB | 1.87 MB | 6.57x | 30.7 KB | 68.8 KB | 292 ms |
| Public Law 118-5 | halves | 2 | 285 KB | 232 KB | 0.81x | 109 KB | 123 KB | 18 ms |
| House Document 110-50 | every 10 pages | 9 | 4.02 MB | 4.13 MB | 1.03x | 38.7 KB | 3.71 MB | 112 ms |
| IRS Form W-9 | every page | 6 | 141 KB | 133 KB | 0.95x | 17.0 KB | 35.6 KB | 122 ms |
| IRS Form 1040 | every page | 2 | 220 KB | 123 KB | 0.56x | 43.4 KB | 79.8 KB | 120 ms |
| Senate Document 105-10 | every page | 67 | 3.06 MB | 3.02 MB | 0.99x | 5.8 KB | 93.2 KB | 522 ms |
| NASA technical report (scan) | every 50 pages | 5 | 12.5 MB | 12.4 MB | 0.99x | 1.51 MB | 3.12 MB | 355 ms |
The two kinds of document behave differently, which the next table shows.
Why parts can be bigger: fonts against scans
A PDF stores a font or an image once and lets many pages use it. When a page is copied into its own file, everything it uses is copied with it. In a text document the pages share a few font files, so every part carries its own copy. A scan has one picture per page and nothing shared, so the parts add up to about the original.
| Kind | Document | Pages | Original | Single pages together | Together vs original |
|---|---|---|---|---|---|
| Text with embedded fonts | Public Law 118-5 | 41 | 285 KB | 1.87 MB | 6.6x |
| Scanned pages | Senate Document 105-10 | 67 | 3.06 MB | 3.02 MB | 1.0x |
| Form | IRS Form 1040 | 2 | 220 KB | 123 KB | 0.6x |
| Form | IRS Form W-9 | 6 | 141 KB | 133 KB | 0.9x |
The law's source file holds 55 streams, 41 of them the pages' contents; one single page of it is 8 streams in its own file, mostly fonts. The Senate document is 2.93 MB of page pictures in 3.06 MB.
Measured: ranges and every N pages
Splitting the law into two halves (pages 1-20 and 21-41) gave 123 KB and 109 KB, 232 KB together from 285 KB. Splitting the 85 page House document every 10 pages gave 9 files totaling 4.13 MB from 4.02 MB (1.03x); the part with the first pages is 3.71 MB, because they include a 3.63 MB picture. Fewer, larger parts repeat fewer fonts.
About the format
A PDF file is a set of numbered objects: pages, fonts, images and the references between them. Operations that copy pages copy the objects they refer to. Wikipedia lists the filters a PDF uses to store data, including FlateDecode (deflate, for text and lossless images) and DCTDecode (JPEG), which is why scanned pages are large and text pages small.
How to choose the parts
- Leave the range empty and set N to split every N pages.
- Type ranges separated by semicolons for named parts: 1-3; 4-10; 11-.
- Use Extract to take one set of pages instead of splitting everything.
- Expect text documents to be bigger in total; use fewer parts if size matters.
Free tools:Extract pages from a PDFDelete pages from a PDFMerge PDF files
Each part holds only its own pages
Copying a page in a PDF can bring the rest of the document along: a form box, a signature field or a link points to other pages, and a copy follows every pointer. We found this in our first tests, where one page of a 41 page law came out with all 41 pages' contents inside. The tool now cuts those pointers and drops anything that can no longer be reached, and we check every part: the number of page objects in the file equals the number of pages shown, and there is exactly one catalog. 145 of 145 outputs across all operations passed.
How it was checked
Each part was opened in pdf.js and compared with the matching pages of the source: page count, the size of every page and the text of every page. All parts passed. Several parts are offered as a ZIP.
Questions people ask
How do I split a PDF?
Choose the PDF, leave the range empty to split every N pages or type ranges such as 1-3; 4-10, and press Split.
Is my PDF uploaded?
No.
Why are the parts together bigger than the original?
In a text document each part carries the fonts its pages use. A 41 page law became 6.6 times its size as single pages. Scanned documents stay about the same.
How can I keep the parts small?
Make fewer, larger parts, or extract only the pages you need.
Does each part contain only its own pages?
Yes. We count the page objects in every part and check that they equal the pages shown.
Does it work on scanned PDFs?
Yes, and the parts add up to about the original: 12.4 MB from 12.5 MB in our 230 page test.
Will the pages look the same?
Yes. Page sizes and text matched in every test.
Can I download all the parts at once?
Yes, as a ZIP.
Does splitting keep fillable forms?
The form boxes are drawn on the pages, but the parts are not fillable forms.
Related
Sources and how we checked
- Wikipedia: PDF (filters) (read 2026-10-10)
- PDF.js (Mozilla): the reader used to check every PDF (read 2026-10-10)
Reviewed October 10, 2026. The measurements ran the tool's own code and then opened every PDF with pdf.js and pdf-lib to compare pages, sizes, text and picture bytes. The files and results are in data/seo/hub/evidence/2026-10-10/pdf.