The attachment is too large: split the PDF instead of squashing it

The message comes back within seconds. “Your message wasn’t delivered because the attachment exceeds the size limit.” Gmail draws the line at 25 MB, Outlook at 20, and a fair number of corporate servers sit lower than either. Upload forms are worse: plenty of them cap out at 10 MB and don’t tell you until you’ve filled in every field.

You have two ways out. Make the file lighter, or make it shorter. Most advice only covers the first one.

When compressing isn’t the answer

Compression works by throwing away detail in the images, and for a lot of documents that’s a fine trade. But sometimes it isn’t. A scanned deed where the notary’s stamp has to stay legible. Site photos in a survey report. A 300-page technical manual where the drawings are the whole point.

There’s also the case where compression simply doesn’t get you there. A 90 MB scan squeezed to 40 MB is still 40 MB, and no amount of squeezing fixes that without turning the pages into mush.

Splitting sidesteps the trade entirely. Every page comes out exactly as it went in, just spread across several files.

Splitting by size instead of by page

	Split by page count	Split by size
What you choose	“Every 20 pages”	“Under 9 MB”
What you get	Parts of wildly different weights	Parts that all fit
Twenty scanned pages	Could be 3 MB or 40 MB	Always under the limit you set
Guessing involved	Yes, and you redo it	None
Good for	A report with chapters	An attachment that keeps bouncing
Where	Split PDF	Split by size

Cutting by weight, not by page count

Cutting by weight, not by page count

Reader has a Split PDF by size tool for this. You give it a ceiling in megabytes and it fills each part up to that ceiling, then starts a new one.

The unit is the page. A page is never cut in half, so a part closes as soon as adding the next page would push it over. That’s why the parts come out uneven: one might hold 40 pages of plain text, the next only 3 because they’re full-bleed scans. The tool isn’t distributing pages evenly, it’s distributing bytes.

The default limit is 25 MB, and there are one-tap presets at 5, 10, 20 and 25. Or type your own number if the form you’re fighting has some other threshold.

The flow

1. Open the tool and drop your PDF on it. It stays on your machine; nothing is uploaded, which is the whole reason the browser does the work here.
2. Set the limit. Pick a preset or type the number in the box.
3. Hit split and wait. Larger documents take a few seconds, because the tool is genuinely building and weighing each part rather than estimating.
4. Look at the table before you download anything. Every part is listed with its page range and its real size in MB.

That table is the part worth reading. It’s the actual measured weight of each file, not a prediction, so if it says 23.4 MB then the file on your disk will be 23.4 MB.

If everything fits in one part, you get a single PDF back. If there’s more than one, you get a ZIP with the parts numbered in order: contract-part-01.pdf, contract-part-02.pdf and so on. Send them in separate emails and say which is which in the subject line, or the person at the other end will open them in whatever order their mail client feels like.

When one page won’t fit on its own

Occasionally a single page weighs more than your whole limit. A 600 dpi colour scan, a full-page aerial photo, a map exported at print resolution. There’s nothing to split there, because pages don’t divide.

That page goes into a file by itself and gets flagged with a warning, in the table and above it, before you download. That warning matters. Handing back a file over the limit without saying so is exactly what makes the second email bounce too. When you see it, run that page through Compress PDF first, or accept that this one page travels a different way.

What doesn’t come along

Being straight about the losses, because they’re real:

• Bookmarks and the document outline don’t survive. The parts are new documents built from copied pages, and the outline belongs to the document, not the pages.
• Form fields stop behaving like a form. They’ll still be visible, but the interactive layer that ties them together doesn’t get carried over. If you need a fillable form intact, don’t split it.
• Document metadata doesn’t travel. Title, author, keywords: the parts start clean. For a lot of people that’s a bonus rather than a loss, and if you’d rather check what was in there first, reviewing a PDF’s hidden metadata takes a minute.
• Password-protected files are refused. Take the password off with Unlock PDF first, then split.

Two more practical things. The parts added together will weigh slightly more than the original, because fonts and images shared across pages have to be embedded again in each part that uses them. And the tool takes files up to 50 MB; above that, compress before you split.

Before you send

Aim under the limit rather than at it. Mail servers add encoding overhead when they attach a file, so a 24.9 MB part can arrive at the gateway as something over 25 and bounce anyway. Set your ceiling at 20 when you’re targeting Gmail and you’ll stop guessing.

And if you want to confirm for yourself that a document this sensitive never left your laptop, open DevTools with F12, watch the Network tab while you split, and see nothing happen. Or pull the wifi and do it offline. There are other ways to check a site isn’t uploading your files, and they’re worth knowing before you trust any tool with a contract.