Split PDF
Type page ranges like 1-3, 5, 7-end — each part becomes its own PDF, downloadable one at a time or all in one .zip. Nothing is uploaded: your file never leaves your device.
Load a PDF first — the field pre-fills with 1-end (the whole document) and the parts rebuild live as you type.
Split parts
Load a PDF and type a range — each comma-separated part is built here live as its own PDF, with its page span, size and a download button. Split into two or more parts and a Download all (.zip) button collects the whole set into one archive.
Each part downloads as its own PDF; two or more parts also get a Download all (.zip) that saves the set in one go, so the browser never asks permission for multiple downloads. Nothing is uploaded either way.
Your PDF never leaves your browser. Pages are copied into the new files in memory on your own device — no server call and no upload. Disconnect from the internet after the page loads and it still works.
Split a PDF into separate files without uploading it
Most "split PDF" sites — iLovePDF, Smallpdf, PDF2Go — work by uploading your file to their server, cutting it there, and handing back a ZIP. For a bank statement, a signed contract or a medical record, that means a copy of a private document sits on someone else's machine. This tool does the same job with zero upload: the split runs in your browser through pdf-lib, a pinned MIT library served from this page. Your file is read locally with the browser's FileReader, each part is assembled in memory, and each is handed straight back as a download. Nothing is transmitted, so there is nothing to leak.
The range syntax: one part per comma
Load a PDF and it immediately shows the page count. Then describe the parts you want — every comma-separated piece becomes its own PDF. On a 10-page document:
| You type | You get |
|---|---|
| 1-3, 5, 7-end | Three PDFs: pages 1–3, page 5, pages 7–10 |
| 1-5, 6-end | Two halves of the document |
| 4 | One single-page PDF of page 4 |
| end | Just the last page |
Ranges are inclusive, whitespace is ignored, and end always means the last page — handy when you don't remember whether the report is 47 or 48 pages. The parts rebuild live on every keystroke; there is no Split button to hunt for. A worked example: a 24-page scanned statement split as 1-8, 9-16, 17-end gives three 8-page PDFs, each small enough to email.
Errors are explicit, never guessed
A splitter that silently "fixes" your input can hand you the wrong pages, and you might only notice after sending the file. So this one refuses instead: 12 on a 10-page PDF is flagged out of range; 5-2 is rejected as reversed with the correct form suggested; 1-3, 2 is an overlap, because a page can belong to only one part; and anything that is not a page or range at all is named in the error. The download buttons only appear when the whole expression is valid.
One click per part — or every part in one .zip
Each part keeps its own download button, and the moment a split produces two or more parts a Download all (.zip) button appears under the list. Two or three parts are fine to click one by one; past that, browsers start interrupting with a permission prompt about downloading multiple files from a site, and the archive removes that friction outright because it is a single save. A 24-page statement split as 1-8, 9-16, 17-end is either three clicks or one 3-entry .zip.
The archive is built here in the tab, by the same first-party writer the SVG to PNG and Image Resizer pages already use — a ~6 KB script served from this site alongside the page, with nothing fetched from anywhere else and no compression library involved at all. It stores the parts uncompressed (ZIP method 0) instead of deflating them, because PDF page streams are already compressed inside the file: deflating them again typically saves a percent or two and can add bytes. Two honest consequences. The .zip is roughly the sum of the parts, not smaller — it is a container, not a compressor. And because nothing is re-encoded, every entry inside is byte-identical to the file that part's own Download button gives you, under exactly the same name. Both routes stay available, and neither uploads anything.
Lossless — and the size caveat that comes with it
Splitting copies whole page objects, so text stays selectable, vectors stay sharp and images are not re-encoded. It does not run OCR on scans, re-flow text, or compress anything — and because every font and image a part's pages use is embedded in full, a part can be larger than its share of the original file. A 10-page extract from a 100-page, 5 MB report is rarely 0.5 MB. If a part needs to be smaller, run it through the PDF Compressor next; to put parts (or other files) back together, use Merge & Split PDF. Password-locked PDFs usually fail to load and are reported, not worked around — unlock them in the app that created them first.