Split PDF Pages

Pulling pages out of a PDF when you cannot tell page 7 from page 7

The page numbered 7 in a PDF is rarely the page printed with a 7 on it. Cover sheets, blank reverse sides, scanned separator pages and annexures somebody inserted three revisions ago all push the count out of step. This is why typing page numbers into a box — the way almost every other splitter works — is guesswork on any document you did not assemble yourself.

This tool renders every page as a numbered thumbnail before it asks you for anything. You click what you can see.

The two selection modes, and when each one is faster

The two modes are not exclusive in spirit, only in mechanism: whichever one the dropdown is set to at the moment you press Extract is the one that supplies the page list. Clicking a thumbnail flips the dropdown to Select pages for you, so the common accident — carefully clicking fifteen thumbnails while the tool is still set to read an empty range box — cannot happen.

Exactly how the range box reads what you type

Most tools document this as “enter page ranges” and leave you to discover the edges by trial. Here is the whole specification.

• Separators are commas. 2, 5, 9 selects three pages.
• A hyphen makes an inclusive range. 4-7 is four pages: 4, 5, 6 and 7. Both ends are included.
• Backwards ranges are corrected, not rejected. 9-4 is read as 4–9. Typing the larger number first is a slip, not an instruction, so it is treated as one.
• Duplicates collapse. 1-5, 3 gives you five pages, not six. Page 3 cannot appear twice in the output no matter how many times you name it.
• Pages past the end are dropped silently. Ask for 1-500 of a twelve-page document and you get all twelve, not an error. This is deliberate: 1-500 is a perfectly reasonable way of saying “everything”.
• Anything unparseable is ignored. A stray semicolon or a word in the middle of the list is skipped; the parts that do make sense are still used.
• Empty results are reported. If nothing you typed matches a real page — 50-60 of a ten-page file — the tool says “That page range does not match any page in this PDF” rather than handing you an empty document. In thumbnail mode the equivalent message is “Select at least one page”.

The one thing worth internalising is the difference between the last two behaviours. Partial nonsense is quietly cleaned up; total nonsense is refused. You will never download a zero-page PDF from this tool, and you will never be blocked by a typo in the eighth of nine range chunks.

Pages always come out in document order, and that is deliberate

Both selection modes sort your page list ascending before anything is copied. Click page 9, then page 2, then page 5, and the result contains pages 2, 5 and 9 in that sequence. Type 9, 2, 5 in the range box and you get precisely the same file.

This is a design decision rather than an oversight, and it is worth being explicit about because it is also a limitation: you cannot use this tool to reorder pages. A document assembled in the order somebody happened to click is almost never what they wanted, and silently producing one would be a trap — the file looks right in the download bar and is wrong when opened. Extraction is one job; sequencing is another.

When reordering is the actual goal, Organize PDF is the tool built for it: drag pages into whatever sequence you want, within a single document.

Your file is never uploaded, and you can verify that yourself

The splitting happens in this browser tab. Your PDF is read from disk into the page’s memory using the browser’s own File API, the pages you chose are copied into a new document there, and the result is written back out as a download. At no point is it transmitted, and there is no server-side fallback for this operation — if something goes wrong you get an error message, not a quiet retry over the network.

Three things follow from that, and they are the practical reasons to care:

• No size ceiling imposed by us. There is no upload limit to run into, because there is no upload. The only real constraint is how much memory your browser will give the tab.
• No copy anywhere but your device. Nothing to retain, nothing to delete on a schedule, no retention policy to take on trust.
• It works offline. Once the page has loaded, disconnect the network entirely. The tool keeps working, which is the most direct proof available that nothing is being sent.

If you would rather check than take our word for it: open your browser’s developer tools, switch to the Network tab, and split a file. You will see no request carrying your document. That test is worth running on any tool that makes this claim, including this one.

Nothing is re-encoded, and what “copying a page” really means

A PDF is not a picture of a document; it is a structured object graph. Pages live in a tree, and each page node points at the resources it needs — fonts, embedded images, vector drawing instructions, colour spaces. This structure is defined by the ISO 32000 standard, and it is what makes clean extraction possible in the first place.

Splitting here copies those page objects, with the resources they reference, into a newly created document. There is no rendering to pixels and no re-compression at any stage. The consequences are worth spelling out, because “no quality loss” is claimed by plenty of tools where it is not true:

• Text stays text. It remains selectable, searchable and copyable in the output, at exactly the fidelity it had in the source.
• Vector graphics stay vector. Charts, logos and line art stay sharp at any zoom, because they are still drawing instructions rather than an image of drawing instructions.
• Scans keep their resolution. A 300 DPI scanned page is still 300 DPI in the extracted file. The image data is carried across untouched.
• Links and annotations come with the pages. A clickable web link on an extracted page still works in the output. A link that pointed at a page you did not extract, though, has nothing left to point at.

There is one exception, and it is the kind of thing splitters normally leave you to discover the hard way. Interactive form fields stop being interactive. A PDF’s fields are registered in a single document-level catalogue, and extraction copies pages rather than that catalogue — so the field keeps its appearance and loses its identity. In practice: whatever was typed into the form is still visible on the extracted page and prints correctly, but the box is no longer clickable, no longer editable, and no longer readable as form data by software that looks for it.

For sending someone a completed form to read, sign or file, that is usually fine and arguably an improvement. If the recipient needs to edit the fields, send them the pages of the original document instead.

The output is normally smaller than the source, simply because it holds fewer pages. It is rarely smaller in proportion, though, and the reason surprises people: a font is embedded once for the whole document, and a font used on even one extracted page still has to be embedded in full. Pull three pages out of a hundred-page report and you might get a file that is a tenth the size rather than a thirtieth. Nothing has gone wrong when that happens.

Password-protected PDFs stop the splitter, on purpose

An encrypted PDF cannot be split until the encryption is removed. The tool refuses rather than attempting to work around it. Run the file through Unlock PDF first — which also runs in your browser — and split the result.

This is the correct behaviour even though it costs you a step. A tool that quietly stripped protection from any document handed to it would be a different and much less trustworthy kind of tool.

Long documents, and what actually costs time

The extraction itself is fast almost regardless of size, because copying page objects is cheap. What takes time on a long PDF is the preview: every page is rendered to a thumbnail, one after another, with no cap on the count. A four-hundred-page document means four hundred renders before the grid is complete, and both the wait and the memory use scale with the page count rather than the file size.

Two practical consequences. First, on a very long document, prefer Enter page ranges — you can type 180-184 while the thumbnails are still filling in. Second, if the tab struggles, it is the preview that is heavy, not the split; closing other tabs genuinely helps, because the constraint is memory available to this one.

Five things this tool deliberately does not do

• It does not produce separate files in one pass. One run, one PDF, containing everything you selected.
• It does not reorder pages. Output is always in document order, as described above.
• It does not delete pages from the original. Your source file is untouched on disk; extraction is a copy operation. To get “everything except pages 4 and 5”, select the pages you want to keep — the result is the same and the original stays intact.
• It does not open encrypted files. Unlock first.
• It does not compress. Fewer pages is the only size reduction on offer here, which is also the only kind that costs no quality.

When you really do need one file per page

For a handful of pages, running the tool once per page is quick — the file stays loaded and the thumbnails stay rendered, so it is a click, extract, click, extract loop rather than a re-upload each time.

For twenty or more, that stops being reasonable, and it is worth being honest that a batch splitter is the right tool for that job and this is not one. What this tool is good at is the far more common case: getting a small, specific set of pages out of a document you need to look at in order to identify them.

Splitting to shrink a file, versus compressing it

When someone hits an attachment size limit, compression is the reflex. It is often the wrong first move.

Compression works by degrading what remains — downsampling images, re-encoding them at lower quality. Splitting removes whole pages at no quality cost to the pages that survive. If the recipient only needs four pages of a forty-page report, sending four pristine pages beats sending forty degraded ones, and it is smaller.

Reach for the PDF Compressor when every page genuinely has to go, and for Compress PDF to 100KB when a hard numeric limit is the problem. Split first, compress second, if you need both.

Where extracting a few pages beats every alternative

The pattern is consistent: a document that exists as one file for filing reasons, and a need for one part of it.

• Pulling a single month out of a year of bank statements, for an application that asked for one month.
• Extracting the signed page of a contract to send on its own, without circulating the other thirty.
• Taking the three slides that matter out of a forty-page deck before a meeting.
• Separating an invoice from the delivery note it was scanned with.
• Removing a cover sheet before uploading a form to a portal that rejects extra pages.
• Splitting a lecture handout into the sections for each week’s reading.

In every one of those cases the pages are identified by looking at them, not by knowing their numbers — which is the entire argument for a thumbnail grid over a text box.

Questions people ask about splitting PDFs

Is my PDF uploaded to a server?

No. It is read from your disk into this browser tab, processed there, and written back out as a download. There is no upload and no server-side fallback for this tool. Disconnect from the network after the page loads and it still works — that is the simplest way to confirm it.

Does splitting reduce the quality of the pages?

No. Page objects are copied with their fonts, images and vector content intact. Nothing is rendered to pixels or re-compressed, so text stays selectable and a 300 DPI scan stays 300 DPI.

Can I get each page as a separate file?

Not in one pass — one run produces one PDF containing your whole selection. Run the tool once per page for a few pages. For a large batch, this is not the right tool.

Can I change the order of the pages?

No. Both modes sort your selection ascending, so pages always appear in document order. Use Organize PDF to rearrange them.

What does the range box accept?

Comma-separated numbers and hyphenated ranges: 1-3, 7, 12-15. Both ends of a range are included, backwards ranges like 9-4 are corrected, duplicates collapse, and page numbers past the end of the document are ignored.

Why did nothing happen when I entered a range?

If no page you named exists, you will see “That page range does not match any page in this PDF.” The usual causes are a range beyond the last page, or the mode dropdown being set to Select pages while the numbers sit unread in the range box.

Why is my extracted file not much smaller than the original?

Fonts are embedded once per document and cannot be partially carried — a font used on a single extracted page is embedded in full. Extracting three pages from a hundred does not produce a file one-thirtieth the size, and that is normal.

My PDF is password-protected and will not split. What now?

Encrypted files are refused rather than worked around. Remove the protection with Unlock PDF first, then split the result.

Is there a file size limit?

No limit is imposed by us, because nothing is uploaded. The practical ceiling is your browser’s memory. Very long documents are slow to preview — every page is rendered as a thumbnail — rather than slow to split.

Do fillable form fields still work after splitting?

No. Anything already filled in stays visible and prints correctly, but the fields themselves stop being editable, because a PDF registers its fields in one document-level catalogue that page extraction does not carry across. Fine for sending a completed form; not what you want if the recipient has to type into it.

Do links on the extracted pages still work?

Web links do — they are carried across with their addresses intact. Links that pointed to a page you did not extract have nothing to jump to any more.

Does the original file change?

Never. Extraction copies pages into a new document and writes a new file. The PDF on your disk is untouched.

What will the downloaded file be called?

Your original name with -pages added, so statement.pdf becomes statement-pages.pdf. Rename it afterwards as you like.

Sources and specifications

• ISO 32000-2 — the current PDF specification, which defines the page tree and object model that make lossless page extraction possible.
• PDF 32000-1:2008 — the freely readable edition of the specification, for the page-object and resource-inheritance behaviour described above.
• PDF Association — PDF 2.0, for plain-language background on what the format guarantees.
• MDN — File API, the browser capability that lets this page read your file without sending it anywhere.
• pdf.js — Mozilla’s PDF engine, which renders the page thumbnails.
• pdf-lib — the library that copies the selected page objects into the new document.

Tools people open next to this one

• Organize PDF — reorder or remove pages inside one document, which is the job this tool deliberately does not do.
• PDF Merger / Splitter — combine several PDFs into one, also entirely in your browser.
• Unlock PDF — remove password protection so an encrypted file can be split.
• PDF Compressor — for when every page has to stay and the file still has to shrink.
• Rotate PDF — fix sideways scanned pages before or after extracting them.
• PDF to JPG Converter — when you need the extracted page as an image rather than a PDF.
• PDF to Word Converter — to edit the text of the pages you pulled out.
• PDF Metadata Viewer — check what a document carries beyond its visible pages.
• All PDF Tools — the full set.