Compress PDF
Make a PDF smaller without making it worse.
or drop files anywhere on this page
Your files stay in this browser. Nothing is uploaded.
About Compress
Compress PDF makes a PDF smaller by re-encoding the images inside it at a lower resolution and quality, so the file is easier to email or upload. Give it a target size, like 200 KB for an exam form or 20 MB for an email, and it finds the sharpest version of your file that fits. It can also convert images to grayscale, drop embedded multimedia and merge duplicated font files, all from the options on this page.
It works image by image, so text stays selectable and vector art stays sharp no matter which quality setting you pick. Page order and page size are never changed. All of this runs in the tab you have open, and the PDF is never sent to a server.
How it works
- Add one or more PDFs. If a file needs a password to open, type it in the field beside the file name first.
- Choose a Target size, such as 200 KB or 1 MB, or Custom size to type your own, and Compress finds the sharpest version under it.
- Or leave Target size on No target and choose an Image quality (Medium, Good or Best) and an Image resolution in ppi yourself. Medium at 72 ppi is the default and gives the smallest files, Best at 300 ppi or higher keeps images crisp.
- Open More options to convert images to grayscale (Image conversion), discard or keep embedded multimedia (Multimedia files), and optionally optimize fonts (Fonts, marked experimental).
- Click Compress PDF. The status line counts through the images as it works.
- Download the result. The result screen shows the size before and after compression.
What Image quality and Image resolution actually do
Image quality is the JPEG setting each image is saved at: Medium is JPEG 65%, Good is JPEG 80% and Best is JPEG 92%. Image resolution is a ceiling rather than a target. The tool takes the largest page width and the largest page height in the document, multiplies each by the ppi you chose, and shrinks any image bigger than the result. On a Letter page, 72 ppi caps an image at 612 by 792 pixels and 144 ppi caps it at 1,224 by 1,584.
Images already under the ceiling keep their pixel size, and nothing is ever enlarged. An image is only replaced when the new version is smaller than the one it came from, so a file that was tightly packed to begin with can come back the same size. Compress from your original file each time, because running a compressed copy through again stacks a second round of JPEG loss on the first.
How Target size picks a setting
Target size uses the same two settings you can set by hand, in eleven steps from gentle to strong: 300 ppi at JPEG 92% first, 72 ppi at JPEG 30% last. It builds the strongest step first, to learn whether the file can fit at all, predicts from that which gentler step should just fit, and then builds that one to check. You get the gentlest step that came in under your size, which usually takes two or three tries.
A size means at most that many thousand bytes: 100 KB is 100,000 bytes and 1 MB is 1,000,000. That keeps the file under the limit whether the site you upload it to counts a kilobyte as 1,000 bytes or as 1,024. File sizes on this site are shown the same way.
The last step, 72 ppi at JPEG 30%, is as far as it goes because that is about where ordinary body text in a scanned page stops being readable; at 50 ppi, text in a phone photo of a page smeared in our tests. When a file cannot fit even there, the result says so, hands you the smallest version it made, and names what is taking the space.
What Compress leaves alone
Compress skips anything it cannot decode safely rather than risk damaging it. In practice that means:
- Text, vector drawings, form fields and page order are not changed.
- Images stored as JPEG 2000, CCITT or JBIG2 are left as they are. CCITT and JBIG2 are the usual formats for a black-and-white scan, so a scan of that kind may not shrink at all.
- Images that are not 8 bits per colour channel, image masks, and pictures drawn inline in a page are skipped too.
- Images with transparency are only re-encoded when they are larger than the resolution ceiling.
- Every image that is re-encoded is saved as an RGB JPEG. CMYK images are converted to RGB, so colours in a print-oriented file can shift slightly.
- The tool sets no file size or page limit. The practical limit is your device's memory, since images are handled one at a time.
Passwords, attachments and fonts
A PDF that needs a password to open shows a password field beside its file name, and the tool will not start until the password is accepted. The compressed copy is saved without the password, so use Protect afterwards if it should stay locked. A PDF that opens freely but restricts printing or copying needs no password, and its compressed copy has no restrictions. The exception, for both kinds of file, is one that could not be made smaller: it comes back untouched, password or restrictions included.
Under More options, Multimedia files is set to Discard by default. That removes video, sound, 3D and other rich-media annotations from the pages, along with the document's list of attached files, and deletes the data they held, so the file shrinks by about that much. An annotation that points at an attached file goes with it, whatever kind of annotation it calls itself; only data the rest of the document genuinely needs, such as a picture a page draws, is kept. Choose Keep if the media has to stay usable.
A worked example: a contract that is too big to email
Say a 30-page contract came off a copier as one full-colour image per page. Leave Image quality on Medium, set Image resolution to 100 ppi and click Compress PDF. Each page image is capped at 850 by 1,100 pixels on a Letter page and saved again at JPEG 65%. If small print in the result has gone soft, start again from the original with Good and 144 ppi.
If the file barely changed, the pages are probably stored in one of the formats above, or the size comes from something other than images. A file that is large because it has too many pages is helped more by Delete Pages or Split by pages. If the pages are scans you want searchable, run OCR first and compress afterwards.
Frequently asked questions
If the file is mostly text and vector graphics, there is little for a compressor to work with, since Compress only re-encodes images. If the images inside it were already efficiently saved, recompressing them would not help, and the original is returned unchanged, which the result screen reports as "same size".
It can, if you choose a low quality and a low ppi together, since both settings control how much detail survives in the re-encoded images. Medium at 72 ppi favours size over sharpness; pick Good or Best and a higher ppi when the images need to stay crisp, such as for printing or zooming in.
It re-encodes the images inside the PDF as smaller JPEGs at the resolution and quality you choose, optionally converts them to grayscale, drops embedded multimedia from the pages, and can de-duplicate repeated font files. Text and vector content are never rasterized.
Yes. Only image data is touched, the text layer is left exactly as it was.
A scan is usually one large image per page, so it has the most to gain from a lower ppi and quality. A document typed from scratch is mostly text, so its size barely moves whichever settings you choose.
By default they are discarded from the document, since embedded multimedia rarely belongs in a file meant to be smaller. The media and attachment data go with them, so the file shrinks by about their size. Open More options and choose Keep under Multimedia files if you need to hold on to them.
Yes. Add multiple PDFs and each one is compressed independently and returned in a ZIP. With a Target size, each file is fitted under that size on its own, and the result lists which ones made it.
It looks for font files that are embedded more than once, byte for byte identical, and keeps a single copy instead of several. It is marked experimental and only affects duplicate font data, not the images or text you see.
Yes, once you have the password. Type it in the field beside the file name. A file that shrinks comes back without the password; one that could not be made smaller comes back exactly as it was, password and all. Use the Protect tool afterwards if the file should stay locked.
Choose Target size, then 20 MB (fits most email). Gmail, Outlook.com and Yahoo Mail allow 25 MB per message, and the Outlook app and iCloud Mail 20 MB. The limit covers every attachment on the message together, so for three PDFs on one email, give each a third of it with Custom size. If a file still does not fit, deleting pages or splitting it into parts will do more than any setting.
Three things set a floor that image settings cannot go below. Pages: in our tests a scanned Letter page still took about 27 KB at the lowest step, so 100 KB holds about three scanned pages. Fonts and text: in a document typed on a computer, the embedded fonts can be most of the file, and Compress does not touch them. And images stored as JPEG 2000, CCITT or JBIG2, common in black-and-white scans, which Compress leaves as they are. The note on the result says which of these it ran into.
Other sizes
Guides
Related tools
This tool ran entirely in your browser and nothing was uploaded. Your Recent files list keeps a copy of local files under 5 MB until you clear it. How your files are handled.