Split PDF: Free Online PDF Cutter
Pull a page range out of a long document, cut a file into parts that fit an upload limit, or separate a report chapter by chapter, then translate only the pages that matter.
Upload or drop document to translate
Max. file size 1 GB
Splitting means different things, so start by naming which one you want
People arrive at a splitter with three quite different jobs in mind. They produce different numbers of files and different outputs, and picking the wrong one is the reason a split has to be redone.
Extract a page range
You want pages 12 to 18 and nothing else. One output file, the rest of the document discarded. This is what most people mean when they say they need to cut a PDF, and it is the job behind pulling one contract clause or one certificate out of a long scan.
Break the file into parts
The whole document is kept, cut into pieces of a fixed length or into one file per page. Nothing is thrown away. This is what you want when a file has to clear an attachment cap, or when a scanned book needs to become individual pages for filing.
Cut along the chapters
The seams follow the document's own bookmark outline rather than a page count, so each chapter, exhibit or section becomes its own file. It needs the source to carry a real outline; without one you are back to counting pages.
Extraction copies pages, it does not cut them out
Pulling a range out of a PDF builds a brand new document and copies the chosen pages into it. Your original is never modified, so there is no risk in trying a range, looking at the result and trying a different one.
Each copied page brings its own resources with it: the fonts it uses, the images placed on it, its annotations and its size box. That is why an extracted page still looks exactly like it did in the long file, and also why a three page extract from a heavily illustrated report is not one hundredth the size of a three hundred page original. Resources are counted per page, not shared out evenly.
One consequence catches people out. If the source document was scanned as images, the extract is scanned images too. Splitting does not turn a picture of text into text, so the pages that come out still need optical character recognition before anything can search or translate them.
What the pieces can no longer reach
Everything a page points at outside itself is a candidate for breaking. A clickable contents page in part one that jumps to a section now sitting in part three has nowhere to land, because the destination page is no longer in the same file. The link is still there and still clickable, and that is what makes it easy to miss.
The bookmark outline has the same problem in reverse. It is a document level object, so each output file needs a trimmed copy that lists only the entries whose pages actually came along. Entries left pointing at pages that are no longer present give a bookmark panel that mostly does nothing.
Then there is a class of reference no software can repair: cross references written into the prose. Legal drafting is full of them. "As set out in clause 14.3" and "see the schedule at page 212" are ordinary text, and once clause 14.3 lives in a different file the reader has to go and find it. When a split document is going to be read by someone else, say in the filename or on a cover page which part of the original each piece came from.
Signed and fillable documents need a second look
A digital signature in a PDF certifies a range of bytes in the file it was applied to. Splitting produces new files with new bytes, so the signature does not travel with the extracted pages and the output is an unsigned document. That is the correct behaviour rather than a defect, since a signature that survived being cut apart would certify nothing. Where a signed original matters, keep it and send the extract alongside it, not instead of it.
Filled forms deserve a check too. The boxes you see are annotations sitting on the page, while the form itself is registered at the document level, so a page taken out of a completed form can arrive with the boxes and without everything that was stored behind them. Open the output and confirm the answers are there before you rely on it. Flattening the form first, which turns the filled values into ordinary page content, removes the question entirely.
Redaction is the opposite trap and a more serious one. Dropping the pages that contain sensitive material does remove those pages, but text that was covered with a black rectangle rather than genuinely removed is still present underneath on every page that comes along. A splitter is not a redaction tool.
Why a long file gets cut up in the first place
- A portal or an email gateway refuses the file for being too large, and it has to arrive in pieces that each clear the cap
- Only a handful of pages are relevant. A single page from a bank statement run, one certificate inside a scanned family file, one clause of a long agreement
- A scanner produced one enormous file from a whole tray of unrelated documents that now have to be separated back out
- Different chapters go to different reviewers or different translators, and each needs their own file
- A records system wants one document per item, so a combined merged file has to be taken apart again
- Only part of a report is public, and the rest must not leave the building
Splitting to translate only what you need
Translation is charged by volume, so the cheapest way through a long document is often to send less of it. A three hundred page technical manual where only the installation chapter matters to the reader is thirty pages of work, not three hundred, once the chapter has been pulled out on its own.
The same trick makes plan limits go further. A DocTranslator subscription covers 100 pages or 30,000 words a month, and the $2 seven day trial covers 10 pages or 3,000 words. Extracting the pages that actually need translating is what lets a large source document fit inside those figures, and it is also the practical way to test quality on a sample before committing a long file.
Once the parts come back translated you can put them together again with Merge PDF, which is a common round trip for reports where only some sections change between editions.
Check the seams, not the middle
When you open the output, look at the first and last page of every part rather than skimming the whole thing. Off by one errors live at the boundaries: a range typed as 12 to 18 when the reader was showing a cover page as page one leaves you one page short at each end. A table or a figure that ran across a page break is the other thing to look for, since it is now half in one file and half in another.
How to split a PDF online
Decide which of the three jobs you are doing
Keeping one range and discarding the rest, breaking the whole file into parts, or cutting along the chapter outline. They produce different numbers of files, so choosing first saves a second attempt.
Note the real page numbers
Work from the page numbers your reader shows, not the ones printed on the page. A cover sheet or a roman numbered preface makes the two disagree, which is where off by one mistakes come from.
Upload the source file
Create a free account, then drag the PDF in or browse to select it. The original is never altered, so you can try a range, look at the output and adjust it.
Open the parts and check the boundaries
Look at the first and last page of each output file, confirm nothing was cut mid table, then translate the piece you actually need into any of 120+ languages.
Split PDF questions
Does splitting change or delete my original file?
No. Extraction builds a new document and copies the selected pages into it, leaving the source exactly as it was. You can try a range, look at what comes out, and try a different one without any risk to the original.
How do I avoid an off by one mistake in the page range?
Use the page numbers your PDF reader shows rather than the numbers printed on the paper. Cover sheets and roman numbered prefaces make the two disagree, so a range that looks right on the printed numbering starts a page or two off. Checking the first and last page of the output catches it immediately.
Can I split a file so each chapter becomes its own document?
That works when the source carries a real bookmark outline, because the outline entries mark where each chapter begins and those are the natural seams. A file with no outline, which includes most scans, gives the tool nothing to follow, so you fall back to page ranges you choose yourself.
What happens to a digital signature when I split a signed PDF?
It does not carry over. A signature certifies a specific range of bytes in the file it was applied to, and a split produces new files with new bytes, so the extracted pages come out unsigned. Keep the signed original and supply the extract alongside it rather than in its place.
Will the links and bookmarks still work in each part?
Anything that points outside its own file cannot work any more. A contents entry in part one that targets a section now in part three has no destination to reach, though the link stays clickable, which is what makes it easy to overlook. Cross references written into the prose, such as a mention of clause 14.3, are plain text and always need a human to follow.
My PDF is a scan. Will the split pages be searchable?
Not by themselves. If the source pages are images of text then the extracted pages are images too, because splitting copies pages rather than reading them. Run optical character recognition on the output, or upload the extract to DocTranslator, which reads text inside scanned pages and images before translating them.
Is splitting a good way to remove confidential pages?
Dropping whole pages does remove them, so that part is sound. What it does not do is deal with sensitive text that was covered with a black box rather than genuinely deleted, since that text is still in the file on every page that comes along. Use a proper redaction tool for that job.
How does splitting help with translation costs?
Translation is priced by volume at $0.005 per word, and plans cover 100 pages or 30,000 words a month, with the $2 seven day trial covering 10 pages or 3,000 words. Sending only the chapter that matters instead of the whole manual is what keeps a large document inside those limits, and it also makes a cheap quality test on a sample possible.
Cut your PDF down to the pages you need
Extract a range, break a long file into parts that clear an upload limit, and translate only the pages that matter instead of paying for the whole document.
Our Partners
More PDF Tools & Translation Services
PDF Tools