Split a PDF into smaller files

Load the document and every page renders as a thumbnail, so you can see exactly what you are cutting rather than trusting page numbers alone. Select the pages you want — click individual thumbnails, or enter ranges like 1-5, 8, 12-14 when you already know the boundaries.

The split runs entirely in the browser: page objects are copied out of the source and written into new PDFs with no re-encoding, so extracted pages keep their original text, vectors, and image resolution perfectly. Download the results individually or as a batch ZIP. A useful sanity check before you close the tab: the pages across all outputs should sum to exactly the source page count.

Watch for two numbering traps. PDFs have a physical page index (the first sheet is page 1) and often a printed page number that starts later because of covers and contents pages — establish which one you are counting before cutting, or your excerpt will open mid-paragraph. Documents with bookmarks give you reliable split points for free: each bookmark marks an intended section start.

For scanned documents without outlines, skim first and note the physical indexes of each section boundary. Doing this once up front beats re-splitting three times because the ranges were off by one.

Extract an invoice from a bank statement so you can expense one page instead of forty. Pull the relevant chapter from a textbook-sized PDF. Break a 200-page scan into single-page files that document-management systems and OCR tools can index properly — many of them choke on large multi-page scans. Or isolate a signature page, sign it, and merge it back afterward.

In every case the pattern is the same: distribute only the pages the recipient needs. Splitting is also a lightweight privacy control — pages you exclude are gone from the output entirely, so fragments can be shared without exposing the rest of the document.

The classic split error comes from mixing two numbering systems. Worked example: a report has a cover, a contents page, and a divider before the body starts at printed page 1 — so printed page 47 is the 50th physical sheet (offset: +3). If someone asks for "pages 47 to 52," add the offset and cut physical 50–55; forget the arithmetic and the excerpt opens three pages late, mid-argument.

Front matter numbered in roman numerals shifts the count the same way — those pages exist physically even though they sit outside the arabic sequence. The habit that prevents all of it: open the file, jump to the requested pages in the viewer, and note their physical positions before entering any range. Then verify after the split: output page counts summed must equal the source count exactly, which catches any range that overlapped or missed.

Password-protected PDFs must be unlocked before they can be split — if your reader opens the file with a password, supply the same password when loading it into the tool; owner-restricted files (copying or printing disallowed) generally need no password at all. Genuinely corrupt files usually trace back to a truncated download, and re-downloading the source fixes them more often than not.

Very large documents are bounded by your device's memory: a thousand-page scan holds every page in the browser tab while processing. If a monster file struggles, extract in chunks of a few hundred pages — the result is identical, since splitting ranges are independent — and download each pass before starting the next. Nothing in this pipeline is re-encoded, so however the file is cut, extracted pages stay pixel-identical to the source.