How to compress a PDF
Add your file by dragging it onto the box or clicking to browse.
Choose a compression level. Lighter settings keep more detail and save less. Heavier settings shrink further and start to soften images.
Compress and download. The saving appears next to the original size, and if it barely moved, the section below explains why.
Your 25 MB email limit is really about 18 MB
This is the reason most people arrive here, and almost nobody is told the actual arithmetic.
Email cannot carry raw binary. Attachments get Base64 encoded for transit, and that encoding inflates them by roughly 33 to 40 percent. The limit your provider advertises applies to the encoded message, not to the file sitting on your desktop.
So Gmail's 25 MB ceiling accepts a file of roughly 18 MB. A 20 MB limit on free Outlook.com takes about 14 MB. The message body, headers, and any other attachments come out of the same budget.
| Provider | Stated limit | Roughly usable |
|---|---|---|
| Gmail | 25 MB | about 18 MB |
| Outlook.com free | 20 MB | about 14 MB |
| Microsoft 365 business | up to 150 MB | set by IT policy, often 25 to 50 MB |
That last row is worth checking rather than assuming. Plenty of organisations cap well below the platform maximum, and the bounce message rarely explains which limit you hit.
Aim comfortably under the number rather than right at it.
Read this before you compress a text document
This tool works by rendering every page to an image and re-encoding it as a JPEG at your chosen quality. That approach is what makes it so effective on scans, and it carries one consequence you need to know about up front.
Your text stops being text. After compression, every page is a picture of a page. You cannot select a sentence, you cannot search the document, and you cannot copy a paragraph out of it. Screen readers lose access to it too.
For a scanned PDF that is no loss at all. It was already images. A phone scan can run 2 to 5 MB per page, so a twenty page scanned contract lands at 50 MB or more, and rasterising it at a sensible resolution routinely cuts 50 to 90 percent.
For a text PDF exported from Word or Google Docs, think twice. Those files are already small, often 300 KB for a fifty page report, because text and font data barely take any space. Rasterising one throws away the searchable text layer and can leave you with a file that is no smaller, or occasionally larger.
Not sure which you have? Try selecting a sentence in your PDF. If individual words highlight, it is a text PDF and you probably do not need this tool. If nothing selects, it is scanned, and this is exactly the right tool.
Either way, keep your original.
Resolution is the lever that matters
The quality slider does two jobs at once. It sets JPEG quality, and it caps the resolution each page renders at. Drag it down and the file shrinks from both directions at the same time.
Resolution is the heavier of the two levers. Pixel count scales with area, so halving resolution quarters the pixels before JPEG quality has even been applied.
For reference: 300 dpi is print quality. Keep it there if the document is going to a professional printer or through OCR later, because character recognition degrades fast on soft scans. 150 dpi is comfortable for screen reading, and on a phone the difference from 300 is close to invisible. Below about 100 dpi, small text starts to break down.
Worth knowing: if your scanner defaults to 600 dpi, as many do, you are producing files four times the size of a 300 dpi scan for no benefit outside archival work. Changing that setting fixes the problem at source better than compressing afterwards ever will.
What compression cannot do
Nothing recovers detail that compression removed. It is one directional, same as a JPG.
Compress a scanned document heavily and small text goes fuzzy at the edges, which matters if someone has to read a reference number off it, or if the file is heading for OCR. Run it through again later and the damage compounds.
It also will not rescue a file that is large for a structural reason. Embedded video, hundreds of embedded fonts, or a CAD export carrying tens of thousands of vector paths all need different treatment.
And if your PDF is already 200 KB, there is nothing here for you.
Frequently asked questions
Why did my PDF barely shrink?
It was probably already a small text PDF. Those carry very little redundancy, so there is not much to remove. The big wins come from scans.
Will the text still be selectable and searchable?
No. Every page is re-rendered as an image, so the text becomes part of that image. If you need a searchable document afterwards, keep the original and send the compressed copy only where searchability does not matter.
What resolution should I aim for?
150 dpi for screen reading, 300 dpi if it is going to print or through OCR. Above 300 is rarely worth the file size unless you are archiving.
Can I compress the same file twice?
You can, but you should not. Losses stack and cannot be undone. Go back to the original and use a heavier setting instead.
How small does a PDF need to be for email?
Smaller than you think. Base64 encoding inflates attachments by a third in transit, so Gmail's 25 MB ceiling really means about 18 MB of file.
Does compressing break a digitally signed PDF?
Yes. Any change to the document structure invalidates the signature. Compress first, then sign.
Related: merge PDFs · split a PDF · PDF to JPG · images to PDF.