How it works
- You add one PDF and type a page range. The file is read into an ArrayBuffer in this tab; the range is validated against the document's real page count before anything is copied, so an impossible range fails immediately rather than producing a broken file.
- pdf-lib creates an empty document and copies only the pages inside your range into it, as page objects. The text layer, embedded fonts and images on those pages come across unchanged — nothing is rasterised and nothing is re-encoded, so the extracted pages are byte-for-byte as sharp as the source.
- Resources shared with pages outside the range are pulled in as dependencies of the pages you kept, which is why a three-page extract from a hundred-page report is not always exactly three hundredths of the size.
- The copy runs in a Web Worker and reports progress per page, so the progress bar reflects real work on a long document rather than an animation.
- The new document is serialised and offered as a blob URL held only by this tab. You download it; closing or reloading the page discards the input and the output.
Limitations
Bookmarks pointing outside the range are dropped. An outline entry is a reference to a page object, and if that page is not in the new document there is nothing for it to point at. Entries that land inside the range are usually preserved, but a nested outline whose parent sits outside the range loses the parent, so the structure flattens. Internal cross-references behave the same way: a 'see page 74' link that resolved by object reference stops working once page 74 is not in the file.
The extract is not always proportionally smaller. If the source embeds a large font subset or a background image used on every page, that resource is copied into the new document because your pages depend on it. Pulling three pages out of a 12 MB report can produce a 3 MB file, and that is correct behaviour rather than a bug.
Splitting does not remove content, it selects it. Anything hidden on a kept page — an OCR text layer under a scan, metadata, an annotation set to invisible — comes with it. If the point of the split is to withhold information, check the extracted pages themselves rather than trusting that dropping the other pages was enough.
Questions
Is there a file size limit?
No cap is enforced. The practical limit is your device's memory, because the whole source document is parsed in the tab before the range is copied. Files over roughly 500 MB can fail on older phones; the same file usually splits without complaint on a laptop.
Why is my three-page extract still several megabytes?
Because those pages depend on resources that were shared across the whole document — most often an embedded font subset or a repeated header image. Those get copied in so the pages still render. Running the extract through Compress PDF will shrink it if the bulk is imagery.
Can I get each page as its own file?
Not in one pass here. Run the tool once per range. If you need a hundred single-page files, that is genuinely a job for a scripted tool rather than a browser tab.
Does splitting reduce the quality of the pages?
No. Pages are copied as objects, not re-rendered. Text stays selectable, vector graphics stay vector, and images keep their original encoding at their original resolution.
Do I need an account?
No. There is no sign-up, no email step and no usage counter. The tool is paid for by the ads on this page.
When people use it
The most common reason to split is that someone asked for one section of something long. A landlord wants the two pages of a bank statement that show the rent debits, not the twelve-page statement. An examiner wants chapter four. A claims form has a field for the invoice and a separate field for the policy schedule, and both arrived in the same PDF.
The second reason is redaction by omission — sending the parts of a document that are relevant and keeping the rest out of the recipient's hands entirely. That works, provided you remember that it removes pages and nothing else. If the pages you are sending contain the sensitive material, splitting will not help; you need a redaction tool that rewrites the content stream.
Splitting is often the first half of a comparison — you pull the same clause out of two versions of a contract and then need to see what moved. Text diff compares the two extracts line by line once you have the wording out of them.
How it compares
Against a server-side splitter, the advantage is not speed — it is that a bank statement never leaves the machine it is sitting on. Against a desktop PDF editor, the honest answer is that the editor wins whenever the job is more than a range: reordering pages, deleting individual pages from the middle, rotating, or splitting one document into twenty at once are all things a browser tab does badly and Acrobat or a command-line tool does in a single operation.