Drop a PDF into the zone below and every page of it appears as a tile you can actually see. Click the scissors between two pages to cut there, or split every few pages, or type the ranges you want. The colored bands tell you which output file each page lands in, and the panel on the right lists the exact documents you are about to download. Nothing is uploaded and the original file is never modified.
Your files are read by JavaScript running on your own machine. There is no upload, no queue, and no copy on our server to delete later.
Drag a card to reorder it, or use the arrows under each card. The order here is the order in the output. Each card is a preview of the finished page, so what you see is what the PDF will contain.
Esc close ← → move R rotate
Starting
This is your processor doing the work, so the time it takes depends on your machine and not on how many other people are on the site.
The splitter above works without reading a word of this. The guide below is for the part that goes wrong later - why four files split from a 40 MB report can add up to more than 40 MB, which split mode matches which real job, what happens to bookmarks and form fields when a document is cut in half, and how to split a 600 page file without your browser tab dying halfway through.
Two minutes, nothing to install, no account needed for your first run. This PDF splitter executes inside the page you are reading, so every step below happens on your own machine.
Click the drop zone above, or drag the file anywhere onto this page. Every page is drawn as a tile. Rendering happens as you scroll rather than all at once, which is why a 600 page document becomes usable in about a second instead of after a minute of waiting.
Click the scissors on a page to cut after it. Or switch to Fixed size and split every 2, 5 or 10 pages, or to Page ranges and type 1-4, 12-30 to build the files you want directly. All three write to the same board. Start with an automatic pass and finish by hand.
The panel on the right names every file you will get, with its page range and page count, and the colored band on each tile matches the file that page belongs to. Read it. If the plan is wrong it is wrong on screen, where fixing it costs one click instead of a whole re-run.
Press Split PDF and the work happens on your processor. Then open the last output file and confirm it starts on the page you expected, because off-by-one errors collect at the end of a document rather than the beginning, which is exactly where nobody looks. Ten seconds well spent.
The output arrives as a single ZIP by default. Switch Delivery to Separate files if you would rather have individual downloads, which is more convenient for two or three files and considerably less convenient for thirty, because most browsers start asking for permission somewhere around the tenth automatic download.
Most online tools that split a PDF ask you to type a page range into a box and press a button. You download a ZIP and find out whether you were right. Usually you were not. If the document was scanned and the printed page numbers do not match the PDF page numbers - the normal case for anything with a cover, a contents page or roman numerals at the front - you find that out three steps too late.
The board above exists to move that discovery forward. Every page is rendered from the actual file, in order, with the position it will have in the output printed in the corner. The colored band along the bottom of each tile is the output file that page ends up in, and it is generated from exactly the same function that decides what gets written when you press the button. There is no second implementation that could disagree with the preview.
That matters more than it sounds. A preview computed separately from the operation is a picture of an intention, and it can be wrong in ways nobody discovers until the ZIP is already open. A preview computed by the operation is a promise. The file list on the right - names, page ranges, page counts - is that same function, printed as text instead of as color.
Page 1 of the file is not always page 1 of the document. A report with a cover sheet, a contents page and two blank separators has its first chapter on PDF page 5. The tiles above are numbered by their position in the file, which is what the splitter uses, so trust the pictures rather than the numbers printed on the paper.
Splitting a document is four genuinely different jobs, which is why there are four modes here rather than one box to type a range into. Pick the wrong one and a five second task becomes a twenty minute one.
| Mode | What it does | The job it is for |
|---|---|---|
| By hand | You click the scissors between pages. Nothing is automatic. | A document with natural boundaries that only a human can see: a bundle of five invoices, a contract with three annexes, a scan of a folder of unrelated letters. |
| Fixed size | Cuts every N pages, straight through. | Chopping a long document into equal chunks for upload limits, printing batches, or a review process where each person gets twenty pages. |
| Page ranges | You type 1-4, 12-30 and get one file per range. | You already know the page numbers. Also the only mode that can leave pages out entirely, and the only one that can put one page into two files. |
| Every page | One file per page, no decisions. | Feeding a document management system, splitting a batch of scanned forms, or building a set of single-sheet PDFs for an archive. |
The other three modes are lossless in the sense that every page of the original ends up in exactly one output file. Page ranges is deliberately not. If you type 1-4, 12-30 on a 40 page document, pages 5 to 11 and 31 to 40 appear in nothing, and the panel on the right says so in plain language before you run anything.
That is the point of the mode. Pulling the summary and the appendix out of a 200 page report and ignoring the 150 pages in the middle is one of the most common reasons anybody opens a splitter at all. The same mode also lets a page appear in two files: type 1, 1-20, 1, 21-40 and the cover sheet is on the front of both halves, which is what you want when the two halves are going to two different people.
If you want the pages you did not select to disappear rather than to be split out, you are looking for Extract Pages or Remove Pages instead. Splitting, extracting and removing sound like the same operation and produce very different files.
Inside the file, a PDF is a graph of numbered objects - fonts, images, color spaces, content streams, annotations - held together by a cross-reference table recording the byte offset each object starts at. A page is one small dictionary pointing at the objects it needs. Nothing is stored in page order. That structure is why splitting is fast, and why the arithmetic on the resulting file sizes surprises people.
When this tool builds an output file, it creates an empty document and copies the page objects you selected into it, following every reference those pages hold and copying the objects on the other end too. A page that draws text in Helvetica pulls the font object across. A page with a photograph pulls the image stream across, byte for byte, still compressed exactly as it was. Nothing is decoded, nothing is re-encoded, and no image on any page is ever touched.
This is the single most surprising thing about splitting a document, and almost nobody explains it. Suppose your 40 MB report embeds a 6 MB corporate font set and a 4 MB logo that appears in the header of every page. In the original, those objects exist once and all 200 pages reference them.
Split it into four files and each file needs its own copy of the font and the logo, because a PDF cannot reference an object that lives in a different file. Four copies of 10 MB of shared resources is 40 MB of duplication that did not exist before. Your four files total 60 MB. None of them is corrupt, wasteful or wrong. Shared resources stop being shared the moment the pages that share them stop living in the same file.
The effect is worst on documents with heavy embedded fonts, a repeated background graphic, or a company letterhead placed as an image on every page. It is negligible on a plain text report or a scan, where each page owns its own content and shares almost nothing.
If total size matters more than page fidelity, split first and then run each output through a compressor, rather than expecting the split itself to shrink anything. Splitting is a structural operation. It moves pages between containers and it never re-encodes what is inside them.
Every online PDF splitter except this one works the same way: your file is posted to a server, processed there, held on disk for somewhere between twenty minutes and an hour, and deleted by a scheduled job you have to take on faith. That architecture is ordinary, well understood, and completely unnecessary for this particular operation.
Splitting a PDF requires no server. It requires a parser and a writer, and both of them run in JavaScript on your own processor. When you drop a file into the zone above, the browser hands the page a reference to bytes that are already on your disk. Those bytes are read into memory, restructured, and handed back to you as a download. There is no request carrying your document, because there is no server-side step for it to depend on.
You can prove it in fifteen seconds without trusting a word of this. Load this page, disconnect from the internet completely, then drag your PDF in and split it. It works. The only network call this page makes during a run is the one that counts your operation against the free allowance, and it carries the name of the tool and nothing else.
Look at why people split a file in the first place. An insurance claim bundle where the loss adjuster gets the photographs and the broker gets the correspondence, a payroll run cut into one payslip per employee, case files separated per client. Different jobs, one shape. The whole document must not reach the same pair of eyes, so a tool that uploads the whole document to achieve that has quietly undone the reason you opened it. If the file you are about to split would be a problem in somebody else's logs, it never enters them.
The trade-off is stated plainly rather than hidden: because the work happens on your machine, a very large document is limited by your device rather than by our hardware. A 300 MB scan will split faster on a laptop than on a six year old phone, and the phone may run out of memory where the laptop does not. A server-side tool would be more consistent there. It would also have your document.
Every tool in this suite carries a section like this one, because a limitation you discover after committing your work to a tool is worth more anger than one stated up front.
The batch limits are 100 MB and 20 files on the free tiers, 500 MB on Premium. Those are the stated numbers. The real constraint is different: a PDF being split is held in memory in its parsed form while the output documents are built alongside it, so peak memory is roughly the source document plus the largest output. A 90 MB file is comfortable on a desktop and marginal on a phone with several other tabs open.
If a very large split fails, it fails cleanly - the operation is refunded and nothing is consumed. Splitting the document in two passes, or closing other tabs first, resolves it almost every time.
| Symptom | Cause | Fix |
|---|---|---|
| The page numbers in the tool do not match the numbers printed on the pages | The document has front matter that is not numbered, or restarts numbering, or uses roman numerals. | Use the tiles rather than the printed numbers. The number in the corner of a tile is always its real position in the file. |
| The output files are bigger than expected | Shared fonts and repeated images had to be duplicated into every file. | Expected behavior, explained in the engineering section above. Compress the outputs afterwards if size matters. |
| Bookmarks are missing from the split files | The outline is dropped deliberately. | Nothing to fix. A contents page built for the whole document would point at pages that are no longer there. |
| The file will not open at all | It is encrypted, or the PDF header is damaged. | Run Unlock PDF for the first case and Repair PDF for the second. |
| The browser blocked the downloads | Separate files delivery triggers several downloads in a row, which browsers throttle. | Switch Delivery back to One ZIP, or allow multiple downloads for this site in the permission prompt. |
| A page appears in two output files | Two typed ranges overlap. | Intentional if you meant it, and the tile badge shows a multiplier when it happens. Adjust the ranges if you did not. |
Ways people actually use a splitter, and the mode that fits each one:
Before you send split files to anybody, open the last one. The first output file is almost always correct, because it starts at page one and there is nothing in front of it to get wrong. The mistakes live at the far end. A range typed one page short quietly drops the final sheet, and the final sheet is usually the one with the signature on it.
Drop the PDF into the zone at the top of this page. Every page appears as a tile. Click the scissors on a page to cut after it, or switch to Fixed size to cut every few pages, or type ranges such as 1-4, 12-30. The panel on the right lists the exact files you will get. Press Split PDF and they download as a ZIP. It is free, there is no watermark, and no account is needed for your first run.
No. The document is read by JavaScript running on your own machine and the split happens on your own processor. Confirm it by loading this page, disconnecting from the internet, and splitting anyway. It works. The only request a run makes counts your operation against the free allowance, and it carries the name of the tool and nothing else - not the file, not its name, not its size.
Because shared resources stop being shared. A PDF stores each font and each repeated image once and lets every page reference it, so when those pages end up in separate files each file needs its own copy, since no PDF can reference an object living in a different file. A document with a 6 MB embedded font set split into four parts carries that font four times. The files are correct, and none of them is corrupt.
No. Page content is copied as compressed streams without ever being decoded, which means the images, the text and the vector artwork in every output file are bit-identical to the original. Splitting rearranges the container. It never re-encodes what is inside it.
Yes. Choose the Every page mode and each page becomes its own PDF, named after the page number it came from. Keep Delivery set to One ZIP on anything longer than about ten pages, because browsers throttle long runs of automatic downloads and start asking permission somewhere around the tenth file.
Scroll to that page on the board and click the scissors on the tile before it. A cut placed after page 12 gives you one file with pages 1 to 12 and another running from 13 to the end of the document. The file list updates immediately. Confirm the boundary there before you run anything.
Not directly. An encrypted PDF cannot have its pages read until the password is removed, so the file is reported as locked the moment it lands, rather than failing once you have already spent an operation on it. Remove the password first with Unlock PDF, on a document you are authorized to open, then split the result.
They are not carried into the output files. A bookmark tree built for a 200 page document points at pages that no longer exist in a 12 page fragment, and copying it would produce navigation that fails silently. Dropping it is the honest behavior. Page content, fonts, page sizes and rotation are all preserved.
Yes, and it is true of every tool rather than this one in particular. A digital signature is a cryptographic statement about one exact sequence of bytes, and splitting produces new files with new byte sequences, so the signature stops validating in all of them. It cannot be worked around. If the signature has to survive, the document cannot be split.
One hundred megabytes per batch on the free tiers and five hundred on Premium. Your device is the real limit though: the document is held parsed in memory while the output files are built alongside it, so peak usage is roughly the source plus the largest output. Desktops cope easily. An older phone with several tabs open may not.
Yes, in Page ranges mode. Any page not covered by a range appears in no output file, and the panel on the right tells you how many pages that applies to before you run. The other three modes always place every page in exactly one output file.
No, on any of the output files. Nothing is stamped on a page, no logo is added, and no trial banner appears on the first sheet of file one, which is exactly where free tools like to put theirs. Each file holds your pages and nothing else. Commercial use is fine and no attribution is required.
Yes. The board is built for touch. Page tiles are large enough to tap accurately, and the scissors sit on the edge of each tile rather than in a gap that would vanish on a narrow screen. Memory is where a phone struggles, not layout, so a 90 MB scan that splits instantly on a laptop can be marginal on a handset with several tabs already open.
Splitting cuts one document into several and normally keeps every page somewhere. Extract Pages takes the pages you pick and gives you just those, leaving the rest behind, while Remove Pages does the mirror image and hands back one document containing everything except the pages you picked. Count the files you want at the end. One file means Extract or Remove, several files means Split.
All of these run the same way: your file is read by JavaScript on your own machine, the work happens on your processor, and the result is handed back to you as a download that never existed anywhere else.
Find the apps that slow your store down, by name and in milliseconds, then answer shoppers' doubts with trust badges under Add to cart. Both are free, with no card and no access to your store admin.
Paste any page and get its three Core Web Vitals, the apps costing time, like Klaviyo or Judge.me, and the Shopify or WooCommerce fix.
Show the promises that settle doubts, from free shipping to a money-back guarantee, right under the Add to cart button. One line in your theme.
Change the link after printing and count the scans. 5 tracked codes on a free account, static codes unlimited.
Anonymous visitors get one operation per 12 hours, per network address. That is not a sales tactic, it is the only lever we have: this tool runs on your processor, not ours, so the run itself earns us nothing. An account tells us who you are, and the ceiling goes up the moment you have one. Your allowance resets in 12 h.
No credit card, no trial period, no sales call. Join with Google, GitHub, Facebook or email.
Free accounts get two operations per 12 hours, and yours reset in 12 h. Waiting costs you nothing and is a completely reasonable choice. If PDF work is a weekly job rather than a one-off, the counter is the wrong tool for you and Premium takes it off the page.
Cancel anytime. Everything you have already produced stays yours either way.