Split a PDF into files under a size limit
Splitting a PDF by file size divides it into sequential parts that each stay under a megabyte limit you choose — 10 MB for a typical email gateway, 25 MB for Gmail, or any custom cap. Pages keep their original order; the tool measures each part by actually saving it as pages are added, so the limit reflects real output bytes rather than an estimate. It handles documents up to 400 pages, entirely in your browser.
How it works
The naive approach — divide 38 MB by 120 pages, conclude each page "weighs" 317 KB, and cut every 31 pages — fails for two structural reasons. First, page weight is wildly uneven: a text-only page can serialize to 3 KB while a full-page scan is 800 KB. Second, PDFs share resources. An embedded font subset or a repeated logo image is stored once and referenced from many pages, so a page's true marginal cost depends on which other pages travel with it: the first page using a font pays for the whole font, the rest pay almost nothing.
The only honest measurement is to build each part and look. The tool works incrementally: it copies pages, in order, into a candidate part and serializes the growing document to bytes, checking the real size as it goes. When appending the next page would push the saved output over your limit, the part is closed one page earlier, downloaded, and a fresh part begins with that page. Serializing includes everything that will be in the final file — copied fonts, images, compressed streams, the cross-reference table — so the number being compared against your cap is the number your email client will see.
Two edge cases are handled explicitly. A single page whose resources alone exceed the limit (say a 12 MB scanned map against a 10 MB cap) cannot be split further — pages are atomic — so it is emitted as its own oversized part and flagged, rather than silently breaking the guarantee. And because shared resources must be duplicated into every part that references them, the parts together are somewhat larger than the source; that overhead is the unavoidable price of making each part independently openable.
The 400-page ceiling exists because of the measurement strategy: repeatedly serializing a growing document is roughly quadratic in the worst case, and 400 pages keeps even image-heavy files responsive on a laptop. If you only need fixed page ranges rather than a size guarantee, the ordinary split-by-range tool has no such cap.
Worked example: a 38 MB survey report through a 10 MB mail gateway
A 120-page property survey — text chapters mixed with photo appendices — weighs 38.2 MB, and the recipient's mail server rejects anything over 10 MB. With the limit set to 10 MB the tool produces five parts:
Part 1: pages 1–41, 9.6 MB (mostly text chapters) · Part 2: pages 42–68, 9.8 MB · Part 3: pages 69–89, 9.4 MB (photo pages, fewer fit) · Part 4: pages 90–108, 9.7 MB · Part 5: pages 109–120, 4.1 MB.
Note the unequal page counts — 41 pages in part 1 but only 21 in part 3 — reflecting where the heavy photographs live. The parts total 42.6 MB, about 11% more than the original, because the report's fonts and recurring header graphics are copied into all five files. Every part sails through the 10 MB gateway, and the recipient reads them in sequence with no missing or reordered pages.
Frequently asked questions
How do I split a PDF into parts under 10 MB for email?
Load the PDF, set the limit to 10 MB (or slightly less to leave room for email encoding), and download the numbered parts. Pages stay in their original order, with each part ending just before it would cross the limit.
Why can't the tool just divide the file size by the page count?
Because pages share resources like fonts and images, and some pages are far heavier than others, arithmetic estimates are unreliable. The tool instead re-saves the growing part after adding pages and reads the true byte count.
What happens if one page is bigger than my size limit?
That page becomes its own part, over the limit, and the tool tells you so — pages are never compressed or cut apart. Run the oversized part through a PDF compressor if it must fit under the cap.
Why is the total size of the parts larger than the original PDF?
A font or image used across many pages is stored once in the original but must be copied into every part that uses it. This duplication typically adds a few percent, more for font-heavy documents.
Is there a page limit when splitting by size?
Yes, 400 pages per input file. The repeated test-saving that makes the size measurement accurate is compute-intensive, and the cap keeps the process fast in the browser.
Does splitting by size upload my document anywhere?
No. The file is split by JavaScript running on your own device, and each part is saved directly from the browser. Nothing is transmitted, which also makes the repeated saving fast.