Splitting a PDF is the one PDF job where the hard part is not the tool. Any tool can cut a document at page 12. Working out that page 12 is where the cut belongs, and that the result should be four files rather than two, is the part that takes the time — and it is the part that goes wrong.

Our Split PDF tool does the cutting in your browser, with nothing uploaded. This is about how to decide what to ask it for.

What "split" actually means here

There are two different jobs people call splitting, and tools are often built for only one of them.

Extracting pages. You want pages 4 to 9 of a long document and nothing else. One input, one output, the rest discarded.

Breaking a document up. A 60-page bank statement PDF is really twelve monthly statements; a scanned batch is really thirty separate invoices that happened to go through the feeder together. One input, many outputs, nothing discarded.

The second is much more common in practice and much less well served, because it needs you to say where every boundary is rather than just one. The ranges box exists for exactly that: each range you type becomes one file.

How to split a PDF in the browser

  1. Open Split PDF and drop in your document. The page count is read immediately, before you type anything, because a range cannot be checked against a document whose length nobody knows yet.
  2. Type the ranges you want. 1-3 is one file of three pages, 8 is page 8 on its own, 12- is page 12 to the end, and a comma starts a new file. The presets write text into the box for you to edit rather than acting on their own.
  3. Read the list of files the tool says it will produce, with the page count of each. Then press Split and take the parts — individually, or all together as a ZIP.

The order is yours. 8, 1-3 produces the same two documents as 1-3, 8, listed the other way round.

Find the boundaries before you type anything

This is the step people skip, and it is where the wasted afternoon comes from.

Open the document and write down the first page of each section. Not the last — the first. Boundaries recorded as starting pages compose into ranges with no arithmetic:

Invoice 1 starts  p.1
Invoice 2 starts  p.3
Invoice 3 starts  p.4
Invoice 4 starts  p.8
Document has      11 pages

1-2, 3, 4-7, 8-

Each range ends one page before the next one starts, and the last one is open-ended so you cannot get the final page count wrong. If you record last pages instead, every boundary needs a mental +1 and the off-by-one errors all land in the middle of the batch where they are hardest to spot.

For a long document, PDF Page Counter will tell you how many pages you are dealing with across a whole folder at once, which is worth knowing before you commit to doing this by hand.

Name the parts so the names survive

A split produces files you will still have next month, so the names matter more than they seem to while you are making them.

Put whatever makes a file findable at the front, and whatever only orders it at the back: 2026-08-northgate-inv-4471.pdf sorts and searches usefully; split-1.pdf, split-2.pdf, split-3.pdf tells you nothing and cannot be sorted into anything meaningful. If the parts are dated, write the date as YYYY-MM-DD — it is the only format that sorts chronologically as text.

If you are splitting a batch you will later want as one document again, keep a note of the original. Merge PDF will put the pieces back together in any order, which makes "split it, fix one part, merge it back" a perfectly reasonable way to correct a document that is mostly right.

A range outside the document is refused, not trimmed

If you ask for pages 12 to 20 of a nine-page PDF, you get an error naming the range rather than a file containing pages 12 to 9.

That is worth understanding rather than working around, because the alternative is the genuinely dangerous behaviour. A tool that quietly clamps the range hands you a document that opens, looks plausible, and is not what you asked for — and there is nothing to check it against. A range written backwards is refused for the same reason.

Overlapping ranges, on the other hand, are allowed: 1-5, 4-9 is a sensible thing to want if two sections share a page.

What can go wrong

A password-protected PDF. A document with an open password cannot be read without it, so it cannot be split. Remove the password in whatever application made the file.

Bookmarks, form fields and signatures. Each part is a new document. Page content is copied faithfully, but document-level extras are not carried across by any splitter, and a digital signature in particular cannot survive being moved into a different file — that is what a signature is for. If a signature matters, keep the signed original.

A very large document. The work holds the source and every output in memory at once, so there is a size cap, and it is a limit of what a browser tab can hold rather than a paywall. For something enormous, split it in two passes.

Scanned pages. Splitting a scan works exactly as splitting anything else does, and the parts are still scans. They have no text layer, so they are not searchable. Extract Text from PDF will tell you in a few seconds whether a document has a text layer at all.

Other ways to do it

On a Mac, Preview can do it: open the thumbnail sidebar, select the pages you want and drag them to the desktop as a new PDF. On Windows, printing to "Microsoft Print to PDF" with a page range set produces an extract, though it re-renders the pages rather than copying them. Adobe Acrobat has a proper split feature if you have a licence.

All of those are fine for one cut. They get tedious at twelve, because each boundary is a separate manual operation rather than a line of text you can read back and correct.

Common questions

Does splitting reduce quality? No. The pages are copied, not re-rendered or re-compressed. Printing to PDF, by contrast, does re-render them.

Can I split by something other than page number — by invoice number, say? Not here. Ranges are page numbers, because that is the only thing a PDF reliably knows about itself. Finding the boundaries is the reading step above.

Can I get one file per page? Yes — list every page as its own range. For a long document, that is a lot of typing, and it is usually worth asking whether one file per page is really what you want.

Can I delete pages instead of extracting them? Same operation, stated the other way round: list the ranges you want to keep. To drop page 7 of an 11-page document, 1-6, 8- is the answer.

Is there a limit on the number of parts? Yes, and the current one is stated on the tool page. It is a browser-memory limit, not a quota.

When splitting is a symptom

Worth noticing: if you are splitting a batch of scanned invoices so that you can open each part, read four figures off it and type them into a spreadsheet, the split is not the task. The typing is.

That is a different job — one row per document under column headings you chose — and it is the one ExtractToExcel does. Converting multiple PDFs into one spreadsheet describes the workflow, and multiple PDFs to one Excel walks through it.