Split a PDF Into Separate Files

How to Split a PDF

1. Select a PDF.
2. Choose page ranges, every N pages, or split into one file per page.
3. Review the output groups before exporting.
4. Download the resulting PDFs (bundled as a ZIP when there is more than one).

What this tool changes

• Produces two or more new PDF files derived from the page ranges you specify.

What it preserves

• Original page content, size, and orientation in each output file.

Good to know

• Very large documents with many split points may take longer to process on low-memory devices.

Which of the three methods you need

By page ranges suits documents whose sections are different lengths — a report where the appendix is four pages and the financials are eleven. You type 1-3,6,9-11 and get one file per group. It is the most work to set up and the only method that handles irregular structure. If you want one of those groups and not the rest of the document as files, pull out specific pages instead — extracting returns a single document holding exactly the pages you name.

Every N pages suits batch scans with a fixed shape: forty pages that are ten four-page forms, split every 4, done in one action. When your document really has that structure this is dramatically faster than typing ranges, and there is nothing to mistype.

Every page as its own file turns forty pages into forty files. Useful for feeding pages one at a time into another system or distributing single sheets to different people, and rarely what you want for documents meant to be read.

Check the arithmetic before splitting every N

The fast method has one failure mode and it is silent. If your forty-page scan is supposed to be ten four-page forms, but one form ran to five pages because someone printed a continuation sheet, splitting every 4 does not error. It produces ten files, every one of them wrong from the fifth form onwards, each holding the tail of one document and the head of the next.

So divide before you split: does the page count go exactly into N? If it does not, the assumption behind the method is already broken and ranges are the honest choice.

If it does divide exactly, spot-check the first page of two or three outputs. The first page of each should look like a first page — a letterhead, a title block, more white space at the top. That check takes fifteen seconds and catches the failure the tool cannot.

What you get back, and what it is called

A split that produces more than one file bundles them into a single ZIP, so you are not saving forty files by hand. Outputs are named after your source with a part number appended — report-part-1.pdf, report-part-2.pdf — and a split producing just one file skips the ZIP and saves directly.

Pages you do not mention end up in no output at all. Splitting does not require you to account for the whole document, which means it is possible to lose pages you meant to keep without being told. If the input was forty pages and your ranges add up to thirty-three, seven pages went nowhere — that may be exactly right, but it should be a decision rather than a discovery. Nor is splitting one-way: if you divide a document further than you meant to you can join the parts back together, though each rebuild clears the document properties again.

Renaming the outputs to something meaningful is a job for your own machine afterwards, and on a large batch it is the slow part of the whole workflow. Splitting into groups you can rename in bulk usually beats splitting into forty individually-meaningful files.

What splitting costs, measured

It is reasonable to assume that splitting a document multiplies it, because each output file has to be a complete PDF in its own right. We measured it on a three-page scan: the three single-page files came to 4,015,102 bytes against 4,013,944 for the same pages in one document. The whole split cost 1,158 bytes, or three hundredths of one percent.

The reason is that every page of a scan owns its image and shares nothing with its neighbours, so separating them duplicates nothing. That also sets the limit of the result: a text document whose pages all draw on one embedded font is the case where the parts genuinely should exceed the whole, since each output file would need its own copy of that font. We have not measured that case, and this number should not be stretched to cover it.

What splitting does cost is everything that belonged to the document rather than to its pages. Each output is a rebuild, so titles, authors, and subjects come back empty, the creator field is replaced with the library name, and bookmarks are gone. On a scanned batch none of that exists to lose; on a long reference document it is the change readers notice first.