Skip to content
WorldofPDFs

Split a PDF into parts of a set size

Break a large PDF into consecutive parts that each stay under a size you set, so every piece clears an attachment limit or an upload cap. Each part is genuinely built and measured rather than estimated from a page count.

Split PDF by Size

measured, not estimated · pages never recompressed

Ready

Loading tool…

How to Split PDF by Size

  1. Add the PDF. Drop in the file. Its size and page count are shown so you know what you are working with.
  2. Set the limit. Type a maximum in kilobytes, or use a preset — 25MB is Gmail's ceiling, and most systems that say 10MB mean it.
  3. Split. Each part is packed with as many consecutive pages as fit under the limit, then the next part starts where it left off.
  4. Download the archive. The parts arrive as one ZIP, because browsers block a page from starting several downloads in a row.

Why the parts have to be built to be measured

The obvious approach — divide the file size by the page count and work out how many pages fit — is wrong, and wrong in a direction that matters. A PDF's size is not the sum of its pages.

Fonts, colour profiles, logos and any repeated image are shared objects referenced from every page that uses them. The first page carrying a font pays for the whole font; the next fifty pages using it cost almost nothing. A scanned page carrying a photograph might be two megabytes while the text page after it is four kilobytes.

So an estimate can be out by an order of magnitude in either direction, and being out in the wrong direction means producing a part that exceeds the limit you set — which is the one thing this tool must not do. Every candidate part is therefore actually assembled and its real size checked.

The search is arranged to keep that affordable: it tries one page, two, four, eight until a part overshoots, then narrows in on the boundary. Each part is found in a handful of builds rather than by testing every page count in turn.

Splitting is not compressing

Nothing here is recompressed. The pages in each part are the pages you started with, at the quality you started with — which is the right behaviour, since a document split for emailing usually still needs to be readable at the far end.

That means the parts add up to slightly more than the original, not less. Each one carries its own copy of the fonts and shared resources the original held once, so ten parts of a font-heavy document can total noticeably more than the single file did.

If the total matters more than the page fidelity, compress first and then split — the parts will be smaller and there will be fewer of them. If a single page is larger than your limit on its own, no amount of splitting will help, and the result says which page it was.

Picking a limit that actually works

Mail systems measure the encoded size of an attachment, not the file on your disk. Base64 encoding adds about a third, so a 25MB Gmail limit is closer to 18MB of actual file. Setting the limit a little below the number in the documentation is worth doing.

Where the recipient's mail server is the tighter constraint — many corporate systems cap at 10MB — that is the number to use, not your own provider's.

For uploads to a portal, check whether the cap applies per file or to the whole submission. Splitting solves the first and makes the second worse.

Frequently asked questions

How do I split a PDF into files of a certain size?

Set a maximum size, and each part is packed with as many consecutive pages as fit under it. The parts come back as a single ZIP.

Will the parts definitely be under my limit?

Yes, unless a single page exceeds it on its own — every part is built and measured before being accepted. Any page too large by itself is named in the result rather than being quietly included.

Does this compress the pages?

No. The pages are copied at their original quality. To make the content itself smaller, run Compress PDF first and then split.

Why do the parts add up to more than the original?

Each part needs its own copy of the fonts and images the original file shared across all its pages. That duplication is the cost of splitting.

Why is it one ZIP rather than several downloads?

Browsers block the second and subsequent automatic download from a page, so delivering the parts individually would silently give you only the first one.

Are my files uploaded?

No. The splitting and measuring happen in your browser and the document never leaves your device.

Why does it take a moment on a long document?

Because each candidate part is genuinely assembled to find out how big it is. That is what makes the limit reliable rather than approximate.

Related tools